University of Southern California

Aug 2021 — May 2023

M.Sc., Machine Learning and Data Science GPA: 3.81 / 4.0

Relevant coursework: Machine Learning, Deep Learning, Databases, Applied and Cloud Computing, Data Structures and Algorithms, Linear Algebra, Probability Theory, Digital Signal Processing.

Vellore Institute of Technology

July 2017 — July 2021

B.Tech., Electronics and Communication Engineering GPA: 8.95 / 10.0

Relevant coursework: Machine Learning, Data Mining and Predictive Analysis, Computer Vision, Big Data Analytics.

Data Scientist

Prime Healthcare

Feb 2024 — Present

• Built a custom MCP server for LLM tool-calling against production databases with guardrails, enabling agent-driven automation across 10+ production workflows and reducing manual data lookup by 70%+.
• Deployed an automated invoice-processing pipeline using vision-language OCR with structured parsing and validation, reducing manual review by 80%+ and replacing an external contracted vendor in production.
• Operationalized a coding-agent stack (Devstral-2, Qwen3-Coder-Next, OpenAI gpt-oss) with tool-calling and standardized evals, improving automation throughput 10x while reducing failure and retry rates in production workflows.
• Delivered a RAG recommendation engine for item substitution with Precision@K, Recall@K, and coverage tracking, improving accuracy by 25%, expanding coverage by 40%, and eliminating 200+ analyst hours/month.
• Architected a multi-model inference platform (vLLM on GPU-backed clustered L40s servers), hosting foundation models and QLoRA domain-adapted variants in parallel with latency and capacity tuning for scalable serving.
• Automated forecasting pipelines for 10K+ SKUs across 51 hospitals, optimizing replenishment under operational constraints and delivering $1.6M/year waste reduction.
• Directed an 8-hospital ERP migration (Ascension-Chicago), implementing mapping and reconciliation checks with cutover runbooks and achieving zero data loss for a $370M+ acquisition.
• Productionized distributed EDI ETL orchestrated with Airflow, defining pipeline SLIs for freshness, failures, and latency; achieved exactly-once delivery with P99 < 5s.
• Scaled contract-audit ETL (3M+ rows/day) via distributed batch processing and async I/O, reducing runtime from 8 hours to 10 minutes and improving reconciliation efficiency by 90% under production load.

Machine Learning Intern

CarmaCam

Aug 2023 — Feb 2024

• Built road-sign detection pipelines for autonomous driving use cases.
• Evaluated AutoML on Google Cloud and transfer learning approaches including ResNet50, Xception, and InceptionResNetV2.

Data Scientist

USC ITS - Office of CISO

Feb 2022 — May 2023

• Redesigned the vendor risk prediction framework for 28,000 USC vendors, reaching an F1-score of 0.91.
• Deployed an XGBoost model with A/B validation and achieved a 15% false-positive reduction.
• Automated vendor risk alerts and reporting in Power BI, saving 20+ hours of manual work weekly.

Data Science Research Intern

Vellore Institute of Technology

Nov 2020 — July 2021

• Developed an efficient deep-learning approach to diagnose COVID-19 and pneumonia from chest X-rays.
• Used U-Net encoder-decoder architectures to double training speed while maintaining low FLOPs.
• Deployed Micronet M3 with 99.3% accuracy and 99.31% F1-score.

Machine Learning Intern

Arista Networks

Nov 2020 — July 2021

• Completed accelerated training in wireless fingerprinting and applied machine learning methods.
• Built indoor positioning models using k-Nearest Neighbor and Random Forest with Wi-Fi and inertial signals, reaching 2-3 meter precision.
• Designed an RSSI-based localization algorithm for real-time client tracking with 0.98 accuracy.