Looking for a job

Adit Jain

@aditjain

Message

I am a PhD ML researcher specializing in reinforcement learning and bandits.

United States

Message

What I'm looking for

I seek a research-focused ML role building interpretable, efficient RL and LLM systems on collaborative teams that value publications, mentorship, and translating research into scalable production impact.

I am a Ph.D. researcher at Cornell University with a 4.0 GPA focused on technical reinforcement learning, stochastic and linear bandits, distributed optimization, and statistical inference, with applied interest in LLMs, retrieval-augmented generation and agentic systems.

I have delivered research internships at Morgan Stanley and Adobe Research where I developed a modified RL algorithm for LLM exploration (benchmarked across reasoning and math environments with modest gains and reduced memory) and an online annotator-sampling algorithm with regret guarantees for high-dimensional sparse bandits. My work includes publications in TMLR, IEEE Access, Control Systems Letters and conference acceptances, and I was a finalist for a top student paper award.

I bring strong theoretical grounding plus practical systems experience (cloud infrastructure, scalable analytics, experiment design) and a broad technical stack. I seek to apply interpretable, efficient ML methods to high-impact problems and to collaborate across research and engineering to move ideas into production.

Experience

Work history, roles, and key accomplishments

PhD Research Intern

Morgan Stanley

Jun 2025 - Aug 2025 (2 months)

Developed a modified RL algorithm for LLMs that uses token-mixture exploration during the thinking phase, improving benchmark performance 5–40% across 15+ reasoning/math environments and reducing memory usage up to 50%.

Reinforcement Learning Large Language Models PyTorch Python Algorithm Design Benchmarking

PhD Research Intern

Adobe Research

May 2024 - Aug 2024 (3 months)

Designed an online sample-selection algorithm that leverages annotator difficulty feedback under a single-expert constraint and proved regret guarantees for an explore-exploit method achieving sublinear regret in high-dimensional sparse bandits despite blocking constraints.

Algorithms Regret Analysis Python Theoretical Proofs Experimental Design Statistical Learning

Education

Degrees, certifications, and relevant coursework

Cornell University

Doctor of Philosophy, Electrical and Computer Engineering

2022 -

Grade: 4.0/4.0

Activities and societies: Research in reinforcement learning, stochastic and linear bandits, and distributed optimization; Data Science Fellowship, Cornell Center for Social Sciences.

Doctor of Philosophy in Electrical and Computer Engineering at Cornell University with a 4.0/4.0 GPA. Advisor: Prof. Vikram Krishnamurthy.

Indian Institute of Technology Guwahati

Bachelor of Technology, Electronics and Communication Engineering (Minor: Computer Science Engineering)

2018 - 2022

Grade: 9.54/10 (major); 9.4/10 (minor)

Activities and societies: Institute Merit Scholarships and departmental top rank; Institute Silver Medalist (listed among achievements).

Bachelor of Technology in Electronics and Communication Engineering with a Minor in Computer Science Engineering, completed 2018–2022. Graduated with GPAs 9.54/10 (major) and 9.4/10 (minor), ranked 1st in respective departmental batches.