Bloomberg Data Science PhD Fellowship 2026: $45k Stipend + Full Tuition

If you’re doing a PhD in AI, machine learning, or data-heavy computer science, this is one of those rare opportunities that can change the entire shape of your early career.

The Bloomberg Data Science Ph.D. Fellowship 2026–2027 doesn’t just throw a small travel grant your way and wish you luck. It pays your full tuition, gives you a $45,000 stipend, pairs you with senior industry researchers, and drops you into a paid 14‑week research internship in New York, London, or Toronto. And if you impress them? You can renew the fellowship for up to three years.

This is Bloomberg quietly building a serious talent pipeline in AI and data science. If you work on large language models, information retrieval, AI for finance, or topics around trustworthy AI, this program is very much aimed at people like you.

Deadline to apply: December 14, 2025. For a competitive application, you can’t treat this like a last‑minute conference submission.

Let’s break it down and then walk through how to give yourself a real shot.

Bloomberg PhD Fellowship at a Glance

What This Fellowship Actually Offers (And Why It’s a Big Deal)

Most “industry fellowships” give you a nice line on your CV and a modest stipend. This one is more like a full‑service package: money, mentorship, and real research time inside a major tech‑finance company.

1. Full tuition coverage

Your university tuition is 100% covered for each year you hold the fellowship. That’s a huge psychological and financial weight off your shoulders. You’re not trying to juggle teaching multiple sections or picking up random side gigs just to keep the lights on. You can plan multi‑year research arcs instead of semester‑by-semester survival.

2. $45,000 stipend you can actually use

On top of tuition, you get a USD $45,000 stipend. This isn’t pocket change.

You can use it for:

• Living costs so you’re not scraping by while doing deep technical research.
• Conference travel (NeurIPS, ICML, ACL, KDD, WWW, you name it).
• Research expenses like GPUs, cloud credits, storage, or specialized hardware.

Think of it as Bloomberg saying, “Do the ambitious version of your PhD project, not the bargain-basement version.”

3. Paid 14‑week, research‑focused internship

Each year you’re a fellow, you’ll complete a 14‑week paid summer research internship at Bloomberg in New York, London, or Toronto.

This is not “make a dashboard and go home” work. They explicitly describe it as research‑focused. You’ll be tackling real problems involving markets, news, search, conversations, and structured financial data.

You’ll get:

• Access to Bloomberg’s huge data sets and production‑grade systems.
• Mentorship from experienced applied researchers and engineers.
• A very clear sense of what it means to do impactful research outside academia.

And yes, if you’re thinking “this sounds like a foot in the door for a future job,” that’s not a bad read.

4. Bloomberg mentors and career support

You’re paired with Bloomberg mentors who guide both your research alignment and your career decisions. That might mean:

• Refining your research topic to better connect with real-world data problems.
• Getting feedback on papers from people who’ve built production systems.
• Strategizing around whether to pursue academia, industry research labs, or hybrid roles.

Mentorship is often the invisible perk that ends up being the most valuable. A 30‑minute call with someone who has solved your exact career dilemma can save you months of spinning your wheels.

5. Up to three years of renewable support

You’re not locked to a single year. The fellowship can be renewed annually for up to three years, at Bloomberg’s discretion. If you keep delivering strong research and maintaining good standing, this can cover you through much of your middle and late PhD years.

Research Topics Bloomberg Actually Cares About

Bloomberg isn’t trying to fund every possible PhD topic. They’re very clear about areas that matter to them, and if you want to be competitive, you should either:

• Work directly in one of these areas, or
• Make a convincing case that your work naturally connects.

Specific topics of interest include, in plain language:

• Agentic AI and LLM Reasoning: Systems where large language models reason, plan, and act over time, not just autocomplete the next token.
• Agent Evaluation and LLM Judges: Methods for having models evaluate other models, or judge outputs, quality, safety, or correctness.
• Code Generation, Semantic Parsing, Text2API: Translating natural language into executable programs, queries, or API calls that actually run and do things.
• Human‑AI Interaction and Data Annotation: How humans collaborate with AI, and better ways to label, curate, and improve data.
• Information Retrieval and Conversational Systems: Search, recommendation, and chat systems that can work well over huge, complex information spaces.
• Knowledge Graphs and Structured Reasoning: Using structured data and explicit relationships to support reasoning and query answering.
• LLM Post‑training, Domain Adaptation, and Alignment: Fine‑tuning, instruction‑tuning, domain specialization (e.g., finance), and safety/alignment work.
• Multi‑modal Models for Document Understanding: Models that handle text, tables, figures, and other document components together—hugely relevant in finance and news.
• Summarization and Content Generation: Automatic summarizing of long texts (reports, earnings, news) or generating high‑quality, controlled content.
• Time‑series Modeling and AI for Finance: Predictive models over time (markets, volatility, risk) and decision‑making in financial contexts.
• Trustworthy AI and Interpretability: Methods to inspect, explain, and trust model behavior—crucial for regulated and high‑stakes domains.

If your topic isn’t literally one of these bullet items but sits nearby (e.g., RL for portfolio optimization; safety‑aware sequence modeling; evaluation frameworks for agent systems), you’re still in play. You just need to draw the line clearly in your proposal.

Who Should Apply (And Who Probably Shouldn’t)

This fellowship is not for people who are merely “interested in data science.” It’s for serious PhD researchers with a technical focus who can plausibly produce influential work.

You’re a strong candidate if:

• You are (or will be) a full‑time PhD student during the 2026–2027 academic year.
• Your expected graduation date is sometime up to the end of 2029.
• Your research area intersects with AI, ML, data science, NLP, IR, or AI for finance.
• You’re comfortable writing for a technical audience and proposing concrete research directions.
• You’re willing and able to spend 14 weeks each summer in New York, London, or Toronto for the internship.

You must remain full‑time for the duration of the fellowship. If you drop to part‑time or leave your program, the remaining support disappears.

Geography-wise, the fellowship is global. That “Africa” tag just indicates that African PhD candidates are encouraged and eligible, not that it’s restricted. The real constraint is visa and work authorization:

• Bloomberg will not sponsor a U.S. visa for this program.
• If you plan to intern in New York, you’ll need self‑sponsored work authorization that covers the full internship/fellowship period (for example, F‑1 OPT or other EAD).
• For London and Toronto, you’ll still need to meet local work authorization requirements; your university’s international office can help you assess feasibility.

You’re not eligible if:

• You’re already supported by another industry fellowship (e.g., a competing big‑tech fellowship).
• You plan on working outside your university during the fellowship (e.g., a part‑time job) without Bloomberg’s explicit permission.
• You can’t commit to the required internship(s) or to attending the kickoff event at Bloomberg HQ in New York each fall (they pay for that, but you must attend unless it’s online-only).

Insider Tips for a Winning Bloomberg Fellowship Application

This fellowship will attract many of the same people who submit to NeurIPS, ICML, ICLR, ACL, KDD, and similar venues. The bar is high. That’s not a reason to back off; it’s a reason to be strategic.

Here’s how to stand out.

1. Treat the research proposal like a mini top‑tier paper

You only get two pages (excluding references) for the research proposal. That means every paragraph has to work hard.

Aim to answer, very clearly:

• What’s the problem?
• Why does it matter in the real world, especially in areas like finance, news, or large‑scale data systems?
• What is your technical approach?
• What are the risks and how will you handle them?
• What are your expected outputs (papers, methods, open source, datasets)?

Think of it as the “Specific Aims” page from an NSF/NIH grant meets the introduction + methodology from a conference paper—compressed and polished.

2. Make Bloomberg’s interests impossible to miss

Don’t just say, “This advances machine learning theory.” Connect the dots for them:

• How could this help better search across millions of documents?
• How might this improve summarization of financial or news content?
• How could this support time‑series forecasting in markets or decision support for investors?
• How does your topic relate to risk, trust, and interpretability in a world full of opaque models?

You’re not writing an advertisement for Bloomberg, but you are showing that your intellectual work and their problems belong in the same conversation.

3. Respect the Overleaf template like it’s a reviewer

They’re very explicit: if you don’t follow the formatting guidelines, your application won’t be reviewed.

That means:

• Use the Overleaf template they provide.
• Don’t mess with margins, font sizes, or spacing to cram in extra text.
• Stay within the page limit.

Losing a shot at this kind of fellowship because you ignored template instructions would be painful, and entirely avoidable.

4. Coordinate early with your advisor

A strong reference letter is not a last‑minute formality here.

Talk to your advisor (or proposed advisor if you’re early in your PhD) at least 4–6 weeks before the deadline. Share:

• Your draft research proposal.
• A short bullet summary of your contributions so far (papers, code, datasets, talks).
• Why you’re interested in Bloomberg specifically.

The best letters don’t just say “X is excellent.” They explain:

• How you compare to other students they’ve worked with.
• Concrete examples of your independence and problem‑solving.
• Why your proposed project is worth funding.

5. Show a track record—or strong signals of potential

If you’re a mid‑PhD student, they’ll want to see:

• Publications or at least submissions to serious venues.
• Open‑source contributions or visible artifacts (libraries, repos, data).
• Evidence that you can finish things, not just start them.

If you’re earlier in your program, they know you might not be stacked with first‑author papers yet. In that case, highlight:

• Strong technical coursework or prior industry research experience.
• Well‑crafted preliminary experiments or prototypes.
• Any TA or mentoring work that shows your depth in the subject.

6. Write for a technical, but not hyper‑narrow, audience

Your reviewers will likely be experts, but not necessarily in your exact sub‑subfield.

That means:

• Define specialized terms briefly the first time you use them.
• Use equations if they genuinely clarify, not to impress.
• Avoid drowning them in acronyms and obscure model variants.

If your advisor in a related area can read your proposal and clearly explain it back to you, you’re in good shape.

Application Timeline: Working Backward from December 14, 2025

Here’s a realistic timeline that doesn’t assume you’re superhuman.

By late July – early August 2025
Start your planning:

• Confirm your eligibility (graduation date, full‑time status, visa feasibility).
• Talk with your advisor about whether this fellowship fits your trajectory.
• Skim recent Bloomberg research talks/publications to understand their interests.

August – September 2025
Draft the core of your research idea:

• Write a 1‑page informal summary of what you want to propose.
• Get feedback from your advisor and one peer.
• Refine your project scope so it’s ambitious but doable in 1–3 years.

October 2025
Move into serious writing:

• Download and set up the Overleaf template.
• Draft the full two‑page research proposal (don’t worry if it’s messy at first).
• Circulate it to your advisor and 1–2 trusted colleagues for comments.

Early–mid November 2025
Polish and prepare supporting materials:

• Tighten the proposal for clarity, flow, and technical precision.
• Finalize your CV focused on research achievements.
• Confirm your referee(s) and give them your final or near‑final proposal plus CV.

Late November – early December 2025
Final checks:

• Create your account on the application portal and fill in all required fields.
• Double‑check formatting, page limits, and file formats.
• Remind your referees of the December 14 deadline and confirm they know how to upload.

By December 12, 2025
Submit everything:

• Aim to submit your own materials at least 48 hours before the deadline.
• Verify in the system that reference letters have been received.
• Keep copies of everything you submitted.

You don’t want to be the person frantically debugging a PDF upload at 23:57 on deadline day.

Required Materials and How to Make Each One Strong

Your application consists of a compact set of documents. Compact does not mean trivial.

You’ll need:

• Curriculum Vitae (CV)
  This should read like a research‑centric CV, not a job resume. Prioritize:

  • Publications and preprints (with clear author order and venues).
  • Talks, posters, and workshops.
  • Open‑source projects (with GitHub links).
  • Awards, scholarships, notable achievements.

• Two‑page research proposal (excluding references)
  This is the heart of your application. Make sure it:

  • Uses the required Overleaf formatting.
  • Clearly states the problem, why it matters, and what’s new.
  • Includes a short but concrete plan: methods, experiments, data.
  • Sketches likely outputs and how you’ll share your work (papers, code, datasets).

• Reference letter from your advisor (required)
  Optionally, you can add a second referee (e.g., a collaborator or previous internship supervisor). During application submission:

  • You’ll enter referees’ contact information.
  • The system invites them to upload letters.
  • Letters must be submitted no later than the application deadline.

  Practically, it’s on you to nag politely and ensure nothing’s missing.

That’s it in terms of components—which is deceptive. With so few moving parts, each one carries heavy weight.

What Makes an Application Stand Out

While Bloomberg doesn’t publish a point‑by‑point rubric, you can infer what matters from the structure of the program.

Expect reviewers to focus on:

1. Research quality and originality
Is the proposed work:

• Novel in a substantive way, not just a small tweak?
• Grounded in relevant literature?
• Likely to produce publishable results in serious venues?

They’ll fund bold but technically credible ideas, not vague dreams.

2. Relevance to Bloomberg’s world

They aren’t a generic tech company. They live at the intersection of:

• Finance and markets
• News and media
• Large‑scale data platforms and search

If your work can plausibly impact any of those—especially through AI, LLMs, time series, IR, or trustworthy AI—that’s a plus. Your proposal should make those connections explicit.

3. Feasibility and focus

Can this actually be done by you, in your situation, in a few years?

• Are your goals scoped to something realistic for a PhD?
• Do you have access to the necessary compute, data, and mentoring?
• Do you acknowledge risks and outline fallback strategies?

Overselling (“We will solve AGI and all of finance in 2 years”) is a fast way to lose credibility.

4. Track record or clear potential

They’ll look at:

• Your prior work (papers, code, projects).
• The strength of your advisor’s letter.
• Signs of independence, curiosity, and resilience.

They are essentially betting on you as a person, not just this one project.

5. Fit with the internship requirement

Your project should be something that benefits from a summer stint inside Bloomberg, not something totally disconnected that you’ll just ignore for 14 weeks every year. Reviewers will favor proposals where the internship and PhD work clearly reinforce each other.

Common Mistakes to Avoid (And How to Fix Them)

A few predictable ways applicants tank their chances:

1. Ignoring the formatting rules

If your proposal doesn’t follow the Overleaf template, it may not even be read. That’s not “harsh”; that’s “they said this upfront.” Fix: download the template on day one and write directly in it.

2. Writing a generic ML research pitch

“Here’s my cool idea for a better transformer” without any nod to Bloomberg’s context is weak. Fix: add at least one concrete use case in finance, news, search, or large structured datasets.

3. Overloading with jargon

If your proposal reads like a stack of paper titles glued together, reviewers will tune out. Fix: write so that a smart ML researcher in another subfield can follow everything without needing to pause and look things up.

4. Late or weak reference letters

A rushed, generic letter can sink you silently. Fix: ask early, share your materials, and give your referees a sense of what the fellowship is looking for.

5. No clear “so what?”

Describing a method without explaining why it matters—scientifically or practically—is a classic error. Fix: explicitly state impact at multiple levels: theoretical, methodological, and application‑level.

Frequently Asked Questions

Do I need to already be working with Bloomberg data or collaborators?
No. There’s no requirement that you’ve previously worked with Bloomberg. What matters is that your research topic is relevant to their world: complex data, AI, LLMs, IR, finance, news, time series, interpretability, etc.

Can I hold this fellowship along with another scholarship or grant?
You cannot hold another industry fellowship at the same time. Non‑industry funding (e.g., government scholarships, university fellowships) might be allowed on a case‑by-case basis. You’ll need to clear that with Bloomberg if you’re in that situation.

What if my graduation is after 2029?
If your expected PhD completion date is after the end of 2029, you don’t meet the stated eligibility window for this round. If you’re on the borderline, talk to your advisor and consider whether your timeline is flexible.

Is this only for students in the U.S.?
No. The program is international. However, Bloomberg will not sponsor a U.S. visa. If you intern in New York, you must already have or obtain self‑sponsored work authorization (such as F‑1 OPT or another form of EAD) that covers the full period. For London and Toronto, you’ll still need to satisfy those countries’ work requirements.

Can I choose which location I intern in (New York, London, Toronto)?
The call specifies that internships take place in one of those three cities. In practice, placement usually depends on project fit, team needs, and your visa/work status. You can express preferences, but treat them as preferences, not guarantees.

What happens if I can’t complete the internship one summer?
The internship is a requirement for each year of the fellowship. If you can’t fulfill it (e.g., visa denial, personal issues), you should discuss directly with Bloomberg, but you should assume your fellowship renewal and ongoing funding could be affected.

Do I need publications to be competitive?
They certainly help, especially in relevant venues. But if you’re early in your PhD, high‑quality ongoing work, strong letters, and a sharp proposal can still make you competitive. Be honest and specific about what you’ve done so far.

Will I get feedback if I’m rejected?
The call page doesn’t promise feedback. Assume that detailed reviewer comments are not guaranteed. That’s another reason to get as much informal pre‑submission feedback as you can from your advisor and peers.

How to Apply and Next Steps

If this fellowship sounds like it fits you, don’t just think “maybe later.” You’re up against people who are already blocking calendar time to work on their proposals.

Here’s what to do now:

1. Confirm basic eligibility
   Check your expected graduation date (must be on or before end of 2029), your full‑time enrollment for 2026–2027, and your potential ability to complete a 14‑week internship in New York, London, or Toronto without needing new visa sponsorship from Bloomberg.

2. Talk to your advisor this month
   Send them a short summary of the fellowship and ask if they’d support an application. If yes, set a shared internal deadline at least 2 weeks before December 14.

3. Draft a one‑page research concept
   Capture the problem, why it matters to Bloomberg’s world, your intended methods, and where it fits into your overall PhD.

4. Move into the official materials

   • Download the Overleaf template for the proposal.
   • Start a serious draft of the 2‑page research statement.
   • Update your CV so it emphasizes research output and potential.

5. Lock in your referees and timeline
   Decide whether you’ll use just your advisor or add an optional second referee. Tell them the exact external deadline and your internal one, and share your draft materials early.

When you’re ready to see the official details and start the application:

Get Started

Ready to apply or want to read the full official description?

Visit the official Bloomberg Data Science Ph.D. Fellowship 2026–2027 page and application portal here:
https://softconf.com/p/Bloomberg2026/

If you’re a serious PhD researcher in AI or data science, this is a tough fellowship to win—but absolutely worth aiming for. Even the process of crafting a proposal at this level will sharpen your thinking and your PhD itself.