I am a Ph.D. Candidate in Computer Science at Georgia Tech, USA, advised by Prof. Vijay Ganesh and Prof. Sriram Vishwanath. I spent one year of my Ph.D. at the University of Waterloo, Canada.

I build AI systems that can reason, verify, and self-correct: grounding the empirical power of large language models (LLMs) in symbolic and formal frameworks, with the long-term goal of systems that are not just capable, but can certify their own correctness.

LLMs are powerful, but their outputs are not inherently verifiable. Software and mathematics offer a different paradigm: compilers, program analysis, symbolic solvers, and proof assistants check correctness with machine-checked guarantees. My research brings these paradigms together across AI for software engineering (AI4SE) and AI for mathematics (AI4Math). At the intersection of automated reasoning and machine learning, I design neuro-symbolic systems in which verifiers and symbolic tools shape how models are trained, how agents search, and what they can prove:

  • Post-training recipes. Reinforcement learning from symbolic feedback and semantic alignment, using compilers and verifiers as training signals (TerraFormer, RLSF, CoTran).
  • Self-improving AI agents. Evolutionary search over entire agent harnesses, so coding agents discover better scaffolding and control flow than hand-designed ones (MILO).
  • Machine-checkable mathematics. Auto-formalization and proof synthesis in Lean, translating informal mathematics into mechanically verified proofs (AIProver, ProofBridge).

My work has appeared at ICLR, ICSE, IJCAI, ECAI, and WWW, including a Distinguished Paper Award at ICSE 2026.

Research Interests Automated Reasoning Neuro-Symbolic AI Coding Agents Formal Methods AI for Math Large Language Models (LLMs)

Previously, I earned my Master's degree in Computer Science from IIT Kharagpur (Institute Silver Medalist, 2022) and my Bachelor's degree in Computer Science from Jadavpur University (University Gold Medalist, 2020), graduating at the top of my class at both the institutes. I have interned twice at AWS AI Labs (Amazon, Seattle) as an Applied Scientist II: in Summer 2026 with the Kiro Science team, leading R&D of MILO, an evolutionary search framework for automated harness discovery and self-improving coding agents, and in Summer 2025 with the Next Gen DevX team, building TerraFormer, a verifier-guided coding agent for Infrastructure-as-Code (ICSE 2026 Distinguished Paper). I am also a recipient of the Reliance Foundation Scholarship for AI & CS (2021), and prestigious national awards JBNSTS, KVPY, and NTSE from the Government of India. Here is my CV.

🏠 I am from the beautiful city Kolkata ('City of Joy'), located in West Bengal, India.

Education

IIT Kharagpur

Kharagpur, India · 2020–2022

M.Tech in Computer Science & Engineering

CGPA: 9.84/10

Work Experience

Amazon Web Services (AWS)

Seattle, USA · May 2026–Present

Applied Scientist II Intern · Kiro Science Team, AWS AI Labs

  • Manager: Purak Jain Mentor: Nikos Kanakaris
  • Leading R&D of MILO, an evolutionary search framework for automated harness discovery and self-improving coding agents via multi-agent orchestration. [arXiv] [Project Page]
  • New SoTA on Terminal-Bench 2.1, PaperBench and DeepSWE; 3 new records on open math problems (EinsteinArena).

Amazon Web Services (AWS)

Seattle, USA · May 2025–Aug 2025

Applied Scientist II Intern · Next Gen DevX (NGDE) Team, AWS AI Labs

  • Manager: Bhavana Bhasker Mentor: Sam Davidson
  • Led R&D of TerraFormer, a neuro-symbolic coding agent for Infrastructure-as-Code; received the highest performance rating (Exceeded Expectations) and a nomination for Georgia Tech's Intern of the Year.
  • Published a first-author paper at ICSE 2026 (CORE A* conference), earning a Distinguished Paper Award.

Latest News

  • Sep 2026 Two papers accepted at NeurIPS 2026 workshops: Learning Flexible Context Management Unlocks Test-Time Scaling at LCFM, and On the Off-Policy Teacher in On-Policy Distillation at MATH-AI and Pre-to-Post. [LCFM] [MATH-AI] [Pre-to-Post]
  • Aug–Dec 2026 Serving as a PC member for AAAI 2027, ICLR 2027, and AISTATS 2027.
  • Aug 2026 Serving as Teaching Assistant for CS 7001: Introduction to Graduate Studies at Georgia Tech (Fall 2026).
  • May 2026 Our paper (ProofBridge) accepted for presentation (non-archival) at the 9th International Symposium on AI Verification (SAIV 2026) and the Workshop on AI for Mathematics & Computer Science Research (AIMACS), both at the Federated Logic Conference (FLoC) 2026 in Lisbon, Portugal, a quadrennial event that brings together leading researchers in logic and computer science.
  • May 2026 Started working as an Applied Scientist II Intern at Amazon Web Services (AWS), Seattle, USA on the Kiro Science Team, for Summer 2026.
  • Apr 2026 Co-organized and spoke at the PLSE/FM + AI Workshop at Georgia Tech (with Google, Oak Ridge National Lab and GTRI), delivering a tutorial on Neuro-Symbolic AI for Code & Math to 150+ Ph.D. students, faculty, and industry professionals. [Site]
  • Apr 2026 Attended ICLR 2026 in Rio de Janeiro, Brazil, where our paper ProofBridge appeared.
  • Mar 2026 Awarded an ICLR 2026 travel grant to attend the 14th International Conference on Learning Representations in person at Rio de Janeiro, Brazil.
  • Jan–Mar 2026 Serving as a PC member for ICML 2026, NeurIPS 2026, and LICS 2026, and as a reviewer for ACM TOPLAS and TMLR.
  • Jan 2026 Paper accepted at ICLR 2026 (acceptance rate: 28%). [arXiv]
  • Dec 2025 Paper accepted at ICSE 2026 (Software Engineering in Practice track) and awarded the Distinguished Paper Award. Work conducted during the Summer 2025 internship at Amazon Web Services (AWS), Seattle, USA. [Link] [arXiv] [YouTube]
  • Nov 2025 Delivered an invited talk on AI for Mathematics at Simon Fraser University, Canada.
  • Oct 2025 Delivered an invited talk on AI for Mathematics at the Fields Institute, University of Toronto, Canada. [YouTube] [Abstract]
  • Aug–Dec 2025 Serving as a PC member for ICLR 2026, AISTATS 2026, and AAAI 2026.
  • Jul 2025 Paper accepted at ECAI 2025 main track (acceptance rate: 23%). [Link] [arXiv]
  • May 2025 Awarded 2nd Place at the College of Computing (CoC) Graduate Poster Symposium at Georgia Tech; recognized for outstanding research presentation at the college-wide, faculty-judged symposium (SCS, SCP, IC, CSE).
  • May 2025 Presented at the High Confidence Software & Systems (HCSS) Conference, Annapolis, MD. [Slides] [Proceedings]
  • May 2025 Started working as an Applied Scientist II Intern at Amazon Web Services (AWS), Seattle, USA for Summer 2025.
  • Jan–Mar 2025 Serving as a PC member for ECAI 2025, NeurIPS 2025, and ICML 2025.
  • Nov 2024 Recognized as a Top Reviewer at NeurIPS 2024, ranking in the top 8% of reviewers; awarded a complimentary full registration.
  • Oct 2024 Selected by ECAI 2024 for the exclusive Lunch with a EurAI Fellow program; had an inspiring discussion on research and career with Prof. Mehdi Dastani (ECAI 2025 Program Chair).
  • Jul–Oct 2024 Serving as a PC member for NeurIPS 2024 (Main & D&B tracks), AAAI 2025, ICLR 2025, and AISTATS 2025.
  • Jul 2024 Paper accepted at ECAI 2024 main track (acceptance rate: 23%). [Link] [arXiv]
  • Jun 2024 Awarded a travel grant to attend IJCAI 2024, Jeju Island, South Korea.
  • Jun 2024 Awarded a full scholarship to attend CAV 2024 (36th Int'l Conference on Computer Aided Verification), Montreal, Canada.
  • May 2024 Short paper accepted at the IJCAI 2024 Doctoral Consortium (acceptance rate: 17%). [Link]
  • May 2024 Serving as Teaching Assistant for CS 3600: Introduction to Artificial Intelligence at Georgia Tech (Summer 2024).

Publications

The full list of papers with venues and PDFs is on the publications page.

Invited Talks

School of Computing Science, Simon Fraser University, Canada

Nov 2025

ProofBridge: Auto-Formalization of Natural Language Proofs in Lean via Joint Embeddings

Mathematical AI Seminar, Fields Institute, UToronto, Canada

Oct 2025

AI for Math: Neuro-Symbolic Auto-Formalization into Lean via Joint Embeddings

Abstract · YouTube

25th High Confidence Software & Systems Conference (HCSS), USA

May 2025

Neuro-Symbolic Techniques for LLM-based Code Generation and Auto-Formalization of Proofs

Slides · Proceedings

ACM Student Chapter, Heritage Institute of Technology, Kolkata, India

Jan 2025

Neuro-Symbolic Approaches for Fine-Tuning Code Language Models

Flyer

Jagadis Bose Scholars Professional Development Forum (JBSPDF)

May 2021

Counterfactual Fairness in Machine Learning

YouTube · Flyer

Teaching Experience

Co-Organizer & Speaker, PLSE/FM + AI Workshop, Georgia Tech · Google · Oak Ridge National Lab · GTRI, USA

Apr 2026

Delivered a tutorial on Neuro-Symbolic AI for Code & Math to 150+ Ph.D. students, faculty, and industry professionals.

Co-Organizer & Instructor, IDEaS One-Day Tutorial on AI4Science, Georgia Tech, USA

Oct 2025

Delivered lectures and hands-on sessions on LLMs and formal methods to 80+ Ph.D. students, postdocs, and scientists.

Teaching Assistant, Georgia Tech, USA

Fall 2026 · Summer 2024

CS 7001: Introduction to Graduate Studies (Profs. Betsy DiSalvo & Kuldeep S. Meel; 60+ students)
CS 3600: Introduction to Artificial Intelligence (Mr. Sourish Dhekane; 120+ students)

Teaching Assistant, University of Waterloo, Canada

Fall 2022 · Spring 2023 · Summer 2023

CS 240: Data Structures and Data Management (Prof. Éric Schost; 400+ students)
CS 105: Introduction to Computer Programming (Dr. Muddassir Malik; 60+ students)
CS 135: Designing Functional Programs (Dr. Byron Weber Becker; 1,200+ students)

Teaching Assistant, IIT Kharagpur, India

Fall 2021

CS 19001: Programming & Data Str. Lab (Drs. Abir Das & Saptarshi Ghosh; 100+ students)
CS 60075: Natural Language Processing (Prof. Sudeshna Sarkar; 120+ students)

Selected Awards & Honors

During Ph.D.

Distinguished Paper Award, ICSE 2026: One of 3 papers recognized among 74 accepted in the SEIP track at the IEEE/ACM 48th Int'l Conference on Software Engineering (2026)

2nd Place, CoC Graduate Poster Symposium, Georgia Tech: Runner-up at the college-wide graduate research symposium across SCS, SCP, IC, and CSE; evaluated by faculty (2025)

Top Reviewer, NeurIPS 2024: Ranked in the top 8% of reviewers (1,300 out of 15,160); awarded complimentary full conference registration (2024)

Lunch with a EurAI Fellow Invitee, ECAI 2024: Invited for a select Ph.D. lunch discussion with Prof. Mehdi Dastani (EurAI Fellow; Chair, ECAI 2025) (2024)

During Master's

Institute Silver Medalist: Highest GPA in M.Tech. (CSE), Rank 1 / 75, IIT Kharagpur (2022)

Best M.Tech. Thesis Award Nominee: Nominated for Best Thesis in M.Tech. Computer Science, IIT Kharagpur (2022)

Reliance Foundation Scholarship (RFS-AI&CS): One of 38 Master's students selected across India; INR 0.6M in academic funding (2021)

CSI Young IT Professional Award: Eastern India Region Runner-Up; youngest participating team in the competition (2021)

GATE Scholar: All-India Rank 188 / 100,000+ (99.81 percentile); stipend of INR 12,400/month for 2 years (2020)

During Bachelor's & High School

University Gold Medalist: Highest GPA in B.E. (CSE), Rank 1 / 70, Jadavpur University (2020)

CSE Dept. Annual Gold Medalist: Highest annual GPA in CSE, Jadavpur University, for three consecutive years (2017–2019)

Special Mention Award, SSCVGIP Computer Vision Summer School: Awarded by the Indian Statistical Institute and IUPRAI for best project (2018)

JBNSTS Senior Scholar: One of 47 scholars selected across India from Science, Engineering, and Medicine (2016)

KVPY Awardee: All-India Rank 143 / 100,000+ (99.86 percentile); fellowship by DST, Govt. of India & IISc Bangalore (2015)

NTSE Scholar: Top 1,000 out of ~1,000,000 examinees; Government of India national scholarship for academic excellence (2012)

Academic Services & Responsibilities

Lead Student Organizer, Foundations of AI Seminar Series (Georgia Tech)

2025–present

Lead organizer for bi-weekly talks hosted by Profs. V. Ganesh, D. Sherrill, and C. Lin; single-handedly manage all logistics: coordinating with visiting faculty and industry researchers, venue, catering, and announcements.

Conference & Journal Services

2020–present

PC member / Reviewer: AAAI (2027, 2026, 2025, 2024, 2023 [AIES]) · ICLR (2027, 2026, 2025, 2024) · ICML (2026, 2025) · AISTATS (2027, 2026, 2025) · NeurIPS (2026, 2025, 2024 [Top Reviewer], 2023, 2021) · LICS (2026) · ECAI (2025) · CAV (2025) · NeSy (2024, 2023) · PLDI (2024) · SAT (2024) · CP (2024) · KDD (2022) · ACM TOPLAS (2026) · TMLR (2026, 2025) · Elsevier ASC (2025) · Nature SR (2024) · IEEE SMC (2023) · IEEE Access (2022) · Springer JCST (2020)

Attended: ICLR 2026 (Brazil) · ICSE 2026 (Brazil) · HCSS 2025 (USA) · NeurIPS 2024 (Canada) · ECAI 2024 (Spain) · IJCAI 2024 (South Korea) · NeurIPS 2023 (USA)

Travel grants: ICLR 2026 (Brazil) · IJCAI 2024 (South Korea) · CAV 2024 (Canada)

IEEE Student Member

2017–present

Member ID: 92577001