Samir Talkal

Architected and delivered an enterprise AI agent in ~3 months, cutting actuarial financial-statement review from 3–4 weeks to ~15 minutes with 92%+ extraction accuracy. Led a 3-engineer team across design, development, deployment, and stakeholder delivery, while implementing Langfuse observability for production monitoring and continuous improvement.

Built an on-prem LLM RAG system with reranking, reducing manual information retrieval time by ~40% and improving response accuracy. Scaled FastAPI microservices to lower latency by ~30% and automated AI-powered ticket creation, reducing manual effort by ~60%.

Reduced infrastructure costs by 30% by building fault-tolerant APIs and deploying on AWS Lambda and API Gateway. Improved multilingual QA latency and request success rate by 10% using LLM-powered Flask chatbots and OpenAI APIs for translation and response optimization.

Enhanced multi-GB data ingestion efficiency by streamlining ETL with Informatica and Airflow for Redshift loads. Reduced pipeline runtime by 15% through improved scheduling and transformation logic, and built Power BI dashboards for real-time operational insights.

Improved classification accuracy by up to 15% across 40 UCR datasets using novel time-series techniques. Boosted model accuracy by 5–15% with SVM and XGBoost using a hybrid feature set (DTW, WEASEL, L1 distance).

Grade: GPA: 3.83 / 4.00

Master of Science in Computer Engineering at New York University (Sep 2022–May 2024), focused on graduate-level computer engineering coursework.

Grade: GPA: 8.62 / 10.00

Bachelor of Technology in Information Technology at Vellore Institute of Technology (Jul 2016–Jun 2020) with an undergraduate focus on IT fundamentals.