Projects

Currently working on (as of 27th January 2026)

LeJEPA-Style Self-Supervised Model for Galaxy Representations

UniverseTBD | Self-Supervised Vision | In Progress (Jan 2026 – Present)

Wunder Challenge 2 — Limit Order Book Forecasting

ML Competition | High-Frequency Time-Series | In Progress (Jan 2026 – Present)

WorldQuant — Macro Risk Exposure Taxonomy from Filings (Exploratory)

WorldQuant | Taxonomy + Prompt Schemas | Jan 2026 - Present  •  Internship memo (PDF)

Quant & Markets

WorldQuant — PIT Relationship Graph Alpha (SEC Filings + Sell-Side Reports)

Quantitative Research Internship | DeepHuman Algo Lab (DeepResearch) | Sep–Dec 2025  •  Internship memo (PDF)

  • Built point-in-time (PIT) pipelines to extract inter-company relationships from 10-K/10-Q SEC Filings (MD&A + Notes sec.) and OCR’d sell-side analyst reports using regex, structured LLM prompting + schema-constrained outputs.
  • Produced directed edge tables with relation classes (supply chain/customer, competitor, litigation, regulatory, M&A, macro/other) and directional AB/BA sentiment signals.
  • Defined an 11-factor edge-weighting ontology (10 directional + 1 direction-invariant) to quantify relationship strength, polarity, and asymmetry—standardized across both data sources.
  • Scaled across 2020–2024 (12k filings, ~400k reports) with QC + provenance fields in under 300USD credits each; packaged graph outputs as backtest-ready tables (indicative internal L/S-neutral results under stated assumptions = ~1.5-1.6 Sharpe from filings, ~2.2-2.3 from analyst reports).

WorldQuant — UPC Shipments Alternative-Data Signals (Issuer-Level Daily Features)

WQ | Aggregation + QC | Sep–Dec 2025  •  Internship memo (PDF)

  • Built issuer-level daily feature tables from UPC shipment streams with point-in-time (PIT) alignment and robust handling of missingness, returns, discontinuations, and vendor breaks.
  • Implemented time-varying UPC weighting & gating using quality/stability signals (e.g., value-share, consistency, co-movement), plus backfills/imputations to stabilize noisy series.
  • Resolved major data-quality and regime issues (corrupted histories, inconsistent mappings, vendor/coverage shifts) with explicit QC flags, repair logic, and audit-friendly provenance.
  • Ran a battery of pre-backtest diagnostics (SNR/strength measures, autocorrelation & stability checks, coverage/turnover analysis, distribution-shift and PIT leakage/sanity checks) to validate signal behavior before downstream backtesting.
  • Delivered backtest-ready aggregates and monitoring/QC fields for downstream research (indicative internal L/S-neutral results under stated assumptions = ~1.3–1.4 Sharpe).

NIFTY 50 Implied-Volatility Reconstruction Challenge

Kaggle Competition | Ranked Top∼75 out of 1500+ participants globally

Monte Carlo Pricing of Temperature Weather Derivatives

Prof. Manabendra Saharia | Course Project

AI/ML Research (LLMs & Multimodal) (publications + projects)

MaCBench — Multimodal Benchmark for Chemistry & Materials Research

M3RG Lab | Workshop (NeurIPS AIforMat Spotlight): OpenReview | Extended version: arXiv:2411.16955 | Journal: Nature Computational Science

  • Co-developed MaCBench, a real-world benchmark spanning data extraction, experimental understanding, and results interpretation for chemistry/materials workflows.
  • Ran systematic evaluations of frontier multimodal models and characterized failure modes beyond basic perception—especially spatial reasoning, cross-modal synthesis, and multi-step inference.
  • Released the initial workshop spotlight paper and contributed to the expanded evaluation + analysis in the extended/journal version.

Knowledge Base Effort

Prof. N.M. Anoop Krishnan & Prof. Mausam | M3RG Lab | GitHub: Repo

  • Developed a knowledge base and data extraction model to aggregate material science data from scientific literature, forming a foundation for domain-specific large language models.
  • Focused on advanced entity linkage and extracting scientific information such as chemical formulas to enhance data accessibility and usability.
  • Built an end-to-end, multi-agent pipeline that ingests publisher XML's of research papers, extracts structured text & tables, mines chemical compositions, and contextually links them to material property queries with evidence & confidence scoring.
  • Productionized a fine-tuned 8B LLaMAT model (local, GPU-aware, low-cost) with domain-adapted prompts, scientific sentence chunking and table classification that feed a growing materials knowledge base.

SmolMoE — Mixture-of-Experts Transformer Language Model (Upcycling + Continued Pretraining)

Cohere Labs — Scholar Take-Home

  • Implemented a compact decoder-only language model with Mixture-of-Experts (MoE) feed-forward blocks, integrating routing logic and MoE-aware training components.
  • Added MoE observability and routing-health monitoring: expert utilization / load distribution tracking, plus a routing-specialization style metric to quantify how selectively experts are used.
  • Built an upcycling path to convert a dense Transformer into an MoE model by copying backbone weights and initializing expert banks from the dense MLP, enabling continued-pretraining from a dense checkpoint.

AstroLLaVA — Astronomy Vision-Language Model

UniverseTBD | Paper: arXiv:2504.08583

  • Contributed to evaluation + deployment workflows for AstroLLaVA, a domain-adapted vision-language model for astronomy built on the LLaVA stack.
  • Supported experimental comparisons on astronomy image–text tasks and helped harden the pipeline for reproducible runs and model release.
  • Project output includes curated astronomy visual QA resources and benchmarking for astronomy-focused multimodal reasoning.

Meta-Agentic Retrieval-Augmented Generation System

Hackathon | Inter-IIT Technical Meet, 2024

  • Developed a platform integrating Retrieval-Augmented Generation (RAG) with dynamic knowledge graph creation, enabling real-time data retrieval from diverse sources like academic papers, web content, and live news.
  • Designed a meta-agent for autonomous optimization, agent creation, and feedback incorporation, enhancing adaptability and accuracy.

Hangman Solver — Multi-Model Ensemble

Trexquant Quant Hiring Challenge | Sequence Models, Boosted Trees & N-gram LMs

  • Models explored: baseline heuristic, BiLSTM consonant predictor, BiLSTM-Attention, LightGBM (26 binary letter classifiers), 1–5 gram frequency tables, static neural–symbolic blends, temperature-scaled fusion, candidate-pruning inference, and an MLP meta-learner (abandoned).
  • System design: length-aware vowel priors, masked-state modeling, calibrated probability fusion, fast n-gram lookup, and constraint-based candidate filtering.
  • Top results: Offline evaluation peaked at 0.81 win-rate (BiLSTM-Attn + calibrated blend + pruning). Public server runs achieved 0.646 mean win-rate (3×1000 games) vs ~18% baseline.

AstroLLaMA

Dr. Ioana Ciuca | UniverseTBD

  • Co-developed AstroLLaMA, a 7-billion-parameter language model fine-tuned from LLaMA-2 using over 300,000 astronomy abstracts from arXiv, achieving a 30% reduction in perplexity compared to LLaMA-2.
  • Facilitated the public release of AstroLLaMA to promote astronomy-focused research, including applications in automatic paper summarization and the development of conversational agents.

AstroTalks

Dr. David Hendriks | UniverseTBD

  • Developed a program leveraging OpenAI's Whisper AI tool to transcribe video datasets accurately.
  • Collaborated with NASA ADS to create and integrate LLM-based pipelines for real-time transcript extraction, topic modeling, and periodic model retraining, enhancing platform performance and accessibility.

Foundations and Engineering (Coursework / Core CS)

Deep Learning for NLP — Language Modeling & Structured Prediction

Prof. Tanmoy Chakraborty | Course Projects

Implementation and Applications of Machine Learning Algorithms

Prof. Rahul Garg | Course Projects

Implementation and Use of Data Structures & Algorithms

Prof. Keerti Choudhary | Course Projects

Truss Analysis

Prof. N.M. Anoop Krishnan | Course Project