cv

M.S. Computer Science

Sacred Heart University, Fairfield, CT

• GPA: 3.8/4.0
• Upsilon Pi Epsilon (UPE) Honor Society

B.Tech Information Technology

GMR Institute of Technology, Vizianagaram, India

• GPA: 8.30/10

Data & ML Engineer

Infosoft Solutions (Remote)

• Optimized ETL pipelines feeding ML feature stores, reducing daily processing latency from 8s to 4.4s (45%) across 10K+ records.
• Trained and deployed a Random Forest classifier on 50K+ user behavioural records achieving 92% accuracy — 23% improvement over prior rule-based system.
• Built model monitoring dashboards in Tableau tracking accuracy drift and data quality metrics, surfacing 35% more upstream discrepancies before production.

Data Analyst

ION Technology Solutions (Remote)

• Built an end-to-end Azure SQL to Python ML pipeline ingesting 300K+ records across 180 tables; lifted revenue prediction R² from 0.44 to 0.80, accurate to within $120.
• Identified 2.3× revenue multipliers on multi-artist events and established commission structure as the single strongest revenue predictor.

CostGuard

• Production-grade LLM reliability proxy with circuit breakers, Prometheus metrics, and $0.000001-precision cost tracking across 12 models and 5 providers. Cuts LLM spend 10–20× vs GPT-4o defaults.

RealDataAgentBench

• Open-source LLM benchmark — 1,180+ runs across 12 frontier models on 39 data science tasks. Surfaced the correctness vs. statistical-validity gap. GPT-4.1 delivers 97% of GPT-5 at 1/15th the cost.

LoRA Fine-tuning of DeepSeek-R1

• Fine-tuned DeepSeek-R1-Distill-Qwen-1.5B with LoRA on GSM8K. 58.2% accuracy on a 1.5B model, 98.8% parameter reduction, 2× inference throughput with Unsloth.