United States

Raghav Mirchandani

Applied Scientist @ Amazon | IIT Delhi

Experience

Trexquant Investment LP

2026 – Present · less than a year

AI Engineering, Machine Learning, Quantitative Research

New York, United States

Amazon

9 years

Lead Applied Scientist

2022 – Present · 4 years · Seattle, Washington, United States

I turn ambiguous business problems in payments, risk, and retail e-commerce domains into shipped ML / Gen AI systems.

Built GenAI Agents / RAG assistants that transform tribal SOPs into searchable workflows and reduced incident diagnosis time ~60%. Shipped recommendation models that lifted CTR ~ 3% while balancing diversity and inventory. Deployed real-time defenses against card enumeration and platform abuse using graph + anomaly detection with 97%+ precision. Behavior-based credit scoring for thin-file customers driving 110K+ new sign-ups in 3 weeks while maintaining default rate guardrails. Improved address understanding with NLP to reduce delivery failures by ~7 - 10% across different global regions (emerging marketplaces IN, LATAM, MENA).

Converting ML signal based treatment levers into measurable KPI metric lifts validated through A/B testing, delivering multi-million dollar impact across revenue and profitability. Mentoring scientists to deliver cutting edge applied research and take work from prototype to production.

Functioning as Senior Machine Learning Engineer, AI engineer, and Research Data Scientist leading and driving high impact, cross functional projects and delivering production scale ML systems. Publishing internal research papers and reviewing for Amazon ML Conference; filed patent for automated root cause triaging; recognized with Amazon Innovation and Cowboy Awards for high-impact contributions.

Applied Scientist

2019 – 2022 · 3 years

Research Scientist

2017 – 2019 · 2 years

NeuralTechSoft

2016 · less than a year

Quantitative Researcher

Algorithmic trading, quant ML research, backtesting and automating multi-strangle option trading strategies earning theta decay while maintaining delta neutrality and lower vega risk. Trained classification models on rolling window volatility, VIX, and momentum to flag high-risk regimes and auto-tune hedging and position sizing. Cut worst-case loss risk ~25% and delivered 15 - 20% annualized returns in backtests; received leadership approval for live trading.

Education

Indian Institute of Technology, Delhi

Integrated M. Tech., Mathematics and Computing · (2012 - 2017)

St. Joseph's Academy, Dehradun

ISC & ICSE · (2012)

Skills & languages

Top skills
Machine Learning Generative AI Recommender Systems
Languages
English (Full Professional) · Hindi (Native or Bilingual) · Sindhi (Elementary)

Get in touch