Experience
My career has been a passion for pushing boundaries and delivering solutions in Machine Learning, AI, and Software Development.
Most recently, during a research co-op, I analyzed reasoning and non-reasoning LLMs on proprietary financial data that gained recognition across the industry—cited by The Wall Street Journal, Barron’s, CNBC, and The Financial Times.
I’m open to full-time roles and collaborations. If you have a project or idea that aligns with my skills and experience, I’d love to connect.
AI Engineer Intern
May 2025 - Aug 2025
United States
Key Responsibilities:
- Led the design and implementation of an end-to-end LLM benchmarking pipeline evaluating 20+ models (OpenAI, Claude, Gemini) on CFA Level III tasks.
- Developed a public, full-stack benchmarking platform (APIs, database, UI) that was featured by the Financial Times and CNBC.
- Defined a rigorous evaluation framework with <2% scoring variance, measuring accuracy, reasoning quality, and token-efficiency/cost.
- Evaluated advanced prompting strategies (zero-shot, CoT with self-consistency, self-discovery) to improve output stability and comparability across models.
ML Engineer Intern
June 2023 - Aug 2023
India
Key Responsibilities:
- Built and deployed a real-time trading signal engine using a custom Java LSTM on Kubernetes, improving throughput by 70% under volatile market conditions.
- Developed a multimodal feature pipeline combining Ichimoku Cloud indicators with FinBERT/GPT-2 sentiment signals, improving prediction accuracy by 15%.
- Improved model performance by replacing ARIMA with LSTM, reducing RMSE from 614.0 to 207.34 and increasing precision by 25%.
- Validated model robustness using ablation studies, rolling-window cross-validation, and early stopping to ensure stability during high volatility.
Knowledge Solutions
ML Research Intern
Sept 2021 - Nov 2021
India
Key Responsibilities:
- Re-architected a distributed Spark ML pipeline, cutting end-to-end runtime by 75% (8+ hours to under 2 hours).
- Improved claim-level risk assessment accuracy by 18% across 500K+ daily claims using ALS and frequent pattern mining.
- Enabled near-real-time scoring by modernizing feature refresh with Spark Structured Streaming and containerized deployment (Docker, Kubernetes).
- Optimized Spark execution by replacing Python UDFs with native functions, using broadcast joins, and applying Adaptive Query Execution to mitigate data skew.
Laugh Out Loud Ventures
SEO Analytics Intern
Feb 2020 - March 2020
India
Key Responsibilities:
- Led a sales team in transitioning to digital advertising during the pandemic
- Optimized Google Ads and Facebook Ads strategies to drive higher conversions for educational courses
- Utilized Google Analytics and CRM tools to refine targeting and improve campaign performance
Open Source Contributor
Jun 2024 - Now
Contribution chart temporarily unavailable.
View GitHub profile