Vishal Fulsundar
Generative AI / Machine Learning Engineer
Mountain view,CA
Overview
6
6
years of professional experience
Work History
Data Scientist / ML Engineer
Xampus
09.2023 - 08.2024
- Boosted subscription conversion by 17% in a live A/B test (n = 50K, p < 0.001, offline Precision@10 = 0.92) by deploying an edge-optimized mentor-student recommender system over 21M profiles on AWS and Docker.
- Slashed customer acquisition costs (CAC) by 30% ( $5K/month saved) while maintaining baseline growth by executing advanced target ad-segmentation analysis on live deployed model data.
- Ensured 100% data reproducibility for downstream model training and analytics by architecting a high-throughput PostgreSQL-to-Parquet ETL pipeline on AWS S3 built with automated data-quality checks.
Graduate Teaching Assistant (Big Data Analytics)
The George Washington University
01.2026 - 05.2026
- Upskilled 42 graduate students in production-level data and ML engineering patterns, Cut manual reporting time by 83% by architecting an AWS Redshift data warehouse with automated data integration pipelines tracking student.
Founder / Operations Manager
Stoffer
03.2020 - 06.2022
- Bootstrapped a B2C marketplace (two published applications) to 220 retailers and 12k customers through rapid MVP iterations, user behavioral cohort analytics, and a data-driven exit decision.
Education
M.S. - Data Science
The George Washington University
05.2026
Skills
- LangChain
- LangGraph
- LlamaIndex
- Hugging Face Transformers
- Multi-Agent Orchestration
- Tool Use
- Function Calling
- Model Context Protocol (MCP)
- Human-in-the-loop workflows
- Document chunking
- Embeddings
- Vector Search
- Qdrant
- FAISS
- Chroma
- Query Rewriting
- Prompt Compression
- Re-ranking
- Top-k retrieval optimization
- Prompt Engineering
- LoRA
- QLoRA
- PyTorch
- Transformers
- Vision-Language Models
- Qwen2-VL
- VLLM
- CNNs
- Scikit-Learn
- XGBoost
- Model Interpretability
- Grad-CAM
- LIME
- SHAP
- Hallucination detection
- DoLa
- Semantic entropy
- Guardrails
- LLM evals
- AUROC
- AUPRC
- ROUGE
- LangSmith tracing
- MLflow
- Statistical Inference
- A/B Testing
- FastAPI
- Docker
- AWS
- Bedrock
- SageMaker
- Lambda
- EC2 g4dn GPU
- S3
- API Gateway
- EventBridge
- Redshift
- GitHub Actions CI/CD
- Streamlit
- Snowflake
- PySpark
- Apache Spark
- ETL Pipelines
- Parquet
- PostgreSQL
- MongoDB
- Neo4j
- Redis
- Python
- SQL
- TypeScript
- JavaScript
- Bash
- R
Selected Genai And Ml Projects
- Full-Stack RAG Document Q&A (WillDocs), LlamaIndex, Qdrant, FastAPI, Docker, Improved top-k retrieval accuracy from 82% to 91% via vector normalization and cosine-similarity re-scoring, while cutting token usage ∼23% through prompt compression, achieving sub-500ms median query latency., Engineered the containerized pipeline behind a FastAPI layer with a modular ingest-to-generation codebase.
- Multi-Agent LLM Orchestration Systems, LangChain, LangGraph, AWS g4dn GPU, Streamlit, Designed an asynchronous 5-agent graph network that decomposes prompts, runs execution paths in parallel over external APIs with resilient error handling, and synthesizes structured markdown results at ∼1.8s end-to-end latency.
Applied Llm Research And Publications
- Hallucination-Detection Reproduction & Evaluation, Implemented and benchmarked three core detectors (DoLa, semantic entropy, layer-semantic dynamics) on Llama-3-8B; engineered a unified AUROC/AUPRC evaluation harness with bootstrap 95% CIs, supported by a 1,250-line CI test suite.
- LLM Output-Stability Study (Neuronpedia), Programmed temperature and token-influence probes across GPT-4 and LLaMA-2; documented systemic repetition collapse patterns past T ≈ 0.7. Manuscript in preparation.
- Publication, Fulsundar, V. "Comparative Study of Extractive and Abstractive Text Summarization (BERT, T5)," JETIR Vol. 10 (2023).
Personal Information
Availability: Open to Relocation
Timeline
Graduate Teaching Assistant (Big Data Analytics)
The George Washington University
01.2026 - 05.2026
Data Scientist / ML Engineer
Xampus
09.2023 - 08.2024
Founder / Operations Manager
Stoffer
03.2020 - 06.2022
M.S. - Data Science
The George Washington University