4+ Years Industry
AI · ML · NLP
Work Experience
From enterprise AI deployments at Fortune 500 companies to cutting-edge LLM research — building systems that move the needle.
Core Competencies
LLMs & GenAI
ML & NLP
MLOps & Infra
Data & Cloud
Work History
University of Wisconsin–Madison
Current
Teaching Assistant — CS 320: Data Science Programming II
- Led hands-on lab sessions focused on Python-based data pipelines, feature processing, and debugging data-centric workflows for large student cohorts.
- Designed and evaluated programming assignments and exams testing data manipulation, feature engineering, and robustness across production-scale datasets.
Auxia
AI Research Engineer (Applied Scientist)
- Built two standalone treatment embedding pipelines (OpenAI + SBERT) with enriched metadata, improving downstream target prediction accuracy by ~12% and enabling client-specific flexibility via Metaflow infra.
- Prototyped behavioral segmentation using LLM-generated user summaries from multi-signal event logs (clicks, views, dwell-time), uncovering latent user groups and boosting CTR prediction accuracy by ~9%.
- Led uplift modeling and causal A/B testing using funnel/binning analysis to identify high-lift segments; applied suppression/force-send rules to optimize personalized targeting.
- Automated and standardized ML training workflows with pipeline orchestration tools, reducing manual coding by 50% and cutting training time by ~25% with cron-scheduled production deployments.
3.7× lift in purchase likelihood
12% accuracy gain
50% less manual coding
~25% faster training
HiLabs
Xtra Miler Award
Machine Learning Engineer II (Senior AI Developer)
- Leveraged diverse generative and encoder LLMs (T5, BioBERT+CRF/LSTM, Mistral, LLaMA) for clinical entity recognition from medical charts, achieving 95% F1 in structured extraction of drugs and diseases.
- Optimized inference via model distillation, layer freezing, and mixed-precision inference, reducing end-to-end extraction time while preserving entity-level precision for safety-critical medical data.
- Led a team of 6 engineers building entity extraction and downstream RAG-based QA systems for healthcare clients using medical charts and Google search results.
- Designed a multi-class webpage classifier (93% precision) using BERT embeddings with XGBoost, incorporating hyperparameter tuning.
- Spearheaded the Provider Directory Accuracy product to improve HEDIS scores for U.S. health plans, driving Fortune 500 adoption and increasing ARR from $11M to $39M.
- Automated web scraping with spaCy NER and proximity search to extract provider data; designed knowledge-graph pipelines and Spark jobs for large-scale analysis.
95% F1 score
$11M → $39M ARR
Fortune 500 adoption
Led team of 6
Publicis Sapient
AI/ML Engineer (Senior Associate, Data Scientist)
- Converted multiple NLP models into containerized microservices deployed on AWS (ECS/EKS), improving scalability and reducing cloud costs by 30%.
- Built a video processing pipeline to ingest media, generate transcripts, and create searchable MCQs — reducing manual content preparation time by 70%.
- Designed an AI-enabled chatbot for a financial client, saving 2,800+ man-hours, reducing support needs by 80%, and cutting case volumes by 4,300+ per quarter; received Certificate of Appreciation.
30% cloud cost reduction
2,800+ man-hours saved
70% less manual work
Certificate of Appreciation
Education
University of Wisconsin–Madison
M.S. Computer Science · Aug 2025 – Dec 2026
Foundation Models · Advanced NLP · Machine Learning · Distributed Systems · Mathematical Principles of RL · Responsible AI · Operating Systems
IIT (BHU) Varanasi
B.Tech Electrical Engineering · Aug 2017 – May 2021
GPA: 8.5/10 · Data Structures & Algorithms · NLP · Probability & Statistics · Linear Algebra · Operations Research
Certifications
🏅 LLM Engineering
☁️ Azure Fundamentals (AZ-900)
🤖 Azure AI Fundamentals (AI-900)
Want the full details? Download my resume tailored to your role.