Chetan Reddy Valluru

AI Researcher

United States2 yrs 3 mos experience

AI EnabledAI ML Practitioner

Key Highlights

1st place at HackRU 2024.
Engineered custom GPT models for chess.
Expertise in RAG, NLP, and edge ML.

Stackforce AI infers this person is a SaaS-focused AI Engineer with expertise in machine learning and data analytics.

Contact

Skills

Core Skills

Agentic AiData AnalyticsMachine Learning

Other Skills

Dockerized microservicesFastAPINL-to-SQL analyticsETL workflowsData migrationPydanticAIMCPIntent classificationTransformer modelsPythonAttention mechanismsETL pipelinesData streamingQuantizationmBERT

About

AI engineer and relentless builder with six years of hands-on experience across undergrad and a Data Science master’s. My passion isn’t just about exploring AI it’s about delivering tangible, scalable impact. I’ve engineered everything from custom GPT models that play chess, to finetuned LLMs for advanced reasoning, and designed agentic AI systems that automate financial research and generate actionable insights and documentation. At JerseySTEM, I shipped vertical AI agents leveraging scalable frameworks like MCP, transforming data access for entire teams. When I see a problem, I build a solution like Taylor.ai, a Chrome extension that instantly generates custom resumes for job seekers, or GemmA.I, an on-device vision assistant for the visually impaired, offering real-time, private scene understanding on iOS. My trajectory is powered by curiosity and execution: 1st place at HackRU 2024, IEEE-published research, and expertise spanning RAG, NLP, edge ML, and multi-agent systems. I thrive where innovation meets impact and I’m determined to create AI that delivers real-world value. If you’re seeking a candidate with raw technical depth, relentless drive, and a proven record of translating ideas into production AI, I’m ready to help you build what’s next.

Experience

2 yrs 3 mos

Total Experience

1 yr 6 mos

Average Tenure

9 mos

Current Experience

Jerseystem

2 roles

AI Engineer

Aug 2025 – Present · 9 mos

Architected and shipped a distributed NL-to-SQL analytics platform using Dockerized microservices and FastAPI, enabling concurrent multi-agent execution and reducing query/data retrieval latency by 40%.
Built a vertical AI operations agent (PydanticAI + MCP) with intent classification/execution planning that routes to Gmail, Drive, Calendar, Docs, read-only MySQL, and schema-aware RAG (Postgres-vector embeddings for meeting transcripts) automating end-to-end workflows (query →artifact →notify/schedule).
Engineered production-grade ETL and data migration workflows across Salesforce/MySQL schemas, including backfills, idempotent loads, and automated validation checks to ensure data integrity and reliable downstream analytics.
Developed high-throughput web crawling and ingestion systems for autonomous agents, optimizing dynamic scraping, caching, retries/backoff, and incremental refresh to sustain continuous data flow.
Productionized LLM deployments with end-to-end observability (token throughput/latency), automated failover, and SLO monitoring, sustaining 99.8% uptime and reliable multi-agent execution.

Agentic AIDockerized microservicesFastAPINL-to-SQL analyticsETL workflowsData migration+1

Artificial Intelligence Engineer

Aug 2025 – Present · 9 mos

Orchestrated containerized NL-to-SQL analytics platform using Docker Compose and FastAPI microservices on AWS Lambda; reduced query latency by 40% via DAG-based request routing with fault-tolerant SQL validation.
Architected Vertical AI Agent with LangGraph DAG orchestration and PydanticAI, managing stateful multi-step workflows across Gmail, Drive, Jira, and Calendar APIs; cut manual process time by 65%.
Deployed production RAG services with Redis-backed context caching and Cohere Rerank on AWS Bedrock; improved retrieval precision from 58% to 84% with OpenTelemetry observability and distributed tracing.
Drove operational excellence via LangSmith agent observability dashboards and on-call monitoring; increased agent reliability by 35% through structured debugging, trace analysis, and GenAI-assisted troubleshooting.

Rutgers university–new brunswick

Research Assistant - Data Science

Nov 2023 – May 2025 · 1 yr 6 mos · New Brunswick, New Jersey, United States · On-site

Trained a decoder-only Transformer from scratch to investigate attention mechanisms; implemented rotary positional embeddings and causal masking to reproduce GPT-3 architecture on academic datasets.
Fine-tuned open-source small language models (Llama/Mistral) on domain-specific corpora, enabling high-fidelity local inference that eliminated dependency on commercial APIs.
Engineered a scalable inference API serving 15k daily predictions for student and faculty workloads, moving experimental models from fragile notebooks to robust shared services.
Designed automated ETL workflows (Python/Airflow) to curate and version large-scale unstructured datasets, ensuring data consistency for reproducible ML benchmarking across research groups.

Transformer modelsPythonETL workflowsMachine Learning

Vnr vignana jyothi institute of engineering and technology (vnrvjiet)

AIML Research Assistant

Aug 2021 – May 2023 · 1 yr 9 mos · Hyderabad, Telangana, India · On-site

Optimized inference infrastructure by implementing 8-bit quantization for mBERT/IndicBERT, reducing memory requirements by 40% while maintaining model fidelity on constrained hardware.
Engineered robust ETL and data streaming pipelines (Python/SQL) with automated drift detection, accelerating research iteration cycles and inference evaluation by 60% across 2M+ documents.
Built a comprehensive evaluation harness for multilingual NLP, establishing rigorous testing protocols that ensured 99.5% operational reliability for deployed advisory models.

ETL pipelinesData streamingQuantizationMachine Learning