Abhinandan Somachetty

About Abhinandan Somachetty

AI/ML Engineer with 5+ years of experience delivering machine learning solutions in finance and e-commerce. Expertise includes predictive modeling, deep learning (Transformers, LSTM, CNN), time-series forecasting, and generative AI (GPT-4, LLaMA). Proficient in MLOps and model deployment using Kubeflow, MLflow, Docker, and Kubernetes. Strong command of cloud platforms (AWS, GCP) and big data technologies (Spark, Hadoop, Kafka) with production experience in SQL/NoSQL databases. Adept at building scalable ML pipelines, conducting model explainability (SHAP, LIME), performing statistical validation, and ensuring compliance in regulated environments. Proven ability to deliver real-time AI systems for trading, credit risk modeling, and recommendation engines with measurable business impact.

Experience

1. Ai Ml Engineer

   JPMorganChase

   Jul 2024 — Present · NY, US

   Designed and deployed predictive models for electronic trading strategies using Python (Pandas, NumPy, TensorFlow) and C++ for low-latency execution, improving intraday trade signal accuracy by 21% across equities and FX desks.• Developed AI-based risk modeling systems leveraging PyTorch, XGBoost, and explainability tools like SHAP and LIME, leading to a 14% reduction in counterparty risk exposure and full compliance with internal governance.• Created time-series forecasting models using Transformers and N-BEATS to predict yield curves, credit spreads, and macroeconomic trends, outperforming ARIMA and Prophet Baselines by 11%.• Led Generative AI initiatives by fine-tuning LLaMA 3 and GPT-4 models for automating financial policy reviews, sentiment analysis, and report summarization; reduced analyst workload by 32%; deployed via AWS SageMaker and EKS.• Engineered and maintained MLOps pipelines using Kubeflow, MLflow, and Docker/Kubernetes, enabling automated model training, validation, and monitoring; reduced deployment cycles by 40% across 15+ production models.• Enhanced real-time feature engineering pipelines using SQL (PostgreSQL) and NoSQL (MongoDB) with indexing and query optimization, cutting latency by 25% for high-frequency trading applications.• Executed A/B testing of trading models, benchmarking legacy statistical approaches versus deep learning architectures (Transformers, LSTM); achieved 8% boost in strategy returns using Bayesian optimization.

Education

• National Institute of Technology Silchar

  B.Tech

  2017 — 2021

• California State University - East Bay

  Master of Science - MS

  2022 — 2024