Open to opportunities

Trayan Das

@trayandas

Message

I build privacy-safe AI and conversational analytics platforms that accelerate enterprise retrieval and reporting.

India

Message

What I'm looking for

I want to build privacy-safe, agentic AI systems end-to-end—text-to-SQL, RAG, and analytics—where I can drive measurable latency and accuracy gains with strong evaluation, observability, and production MLOps.

I’m an AI/ML Senior Analyst currently engineering end-to-end, privacy-safe conversational analytics platforms. I architect privacy redaction (PII/profanity), hybrid BM25 + vector AI search across 100k+ documents, and multi-agent workflows—aiming to cut retrieval latency and make analysis faster for real teams.

In my recent work, I designed a text-to-SQL agent chain that turns natural-language questions into T-SQL and executes directly against Lakehouse SQL endpoints. This enables analytics over 10M+ records in under 30 seconds, while pairing agent orchestration with practical reliability patterns for enterprise adoption.

I also build agentic RAG systems for context-aware business document generation using LangGraph and LoRA fine-tuned LLMs, reducing drafting time by 60–70%. Earlier, I delivered anomaly detection with PyTorch LSTM autoencoders (improving risk detection ~20%) and KPI pipelines with PySpark/Databricks (reducing manual reporting ~50%), and I’ve implemented predictive maintenance models that improved recall (~85%) and reduced downtime (~30%). I’m driven by measurable performance gains, strong evaluation, and production-minded MLOps practices.

Experience

Work history, roles, and key accomplishments

Current

Senior AI/ML Sr. Analyst

Current

Accenture

Sep 2024 - Present (1 year 10 months)

Architected a privacy-safe conversational analytics platform on Azure with PII/profanity redaction, hybrid BM25 + vector search over 100k+ documents, and a multi-agent system to cut retrieval latency. Designed text-to-SQL and agentic multi-agent RAG workflows for automated business document generation on Azure.

Azure PII and Profanity Redaction Incremental Ingestion BM25 Vector Search Multi Agent Systems Text To SQL T SQL Lakehouse SQL LangGraph RAG CoPilot Studio Azure AI Foundry Azure OpenAI

Data Scientist

Expro

Sep 2022 - Sep 2024 (2 years)

Built a PyTorch LSTM-autoencoder RAG-based anomaly detection approach for well-sensor time-series to improve risk detection and reduce analysis time. Developed a PySpark KPI pipeline on Azure ML/Databricks using PCA, K-means, and Random Forest to derive telemetry-based KPI features.

PyTorch RAG Anomaly Detection Hugging Face Transformers Pyspark Azure Machine Learning Databricks K Means Random Forest scikit learn

Data Scientist

Ganit Labs

Mar 2021 - Sep 2022 (1 year 6 months)

Implemented predictive maintenance models on GCP for oilfield equipment, achieving strong recall and reduced downtime. Worked with GCP services including BigQuery, Cloud SQL, Cloud Storage, and App Engine in the model lifecycle.

GCP Predictive Maintenance BigQuery Cloud SQL Cloud Storage

Biomedical Engineer

Portea

Aug 2018 - Sep 2020 (2 years 1 month)

Led cross-functional medical equipment maintenance and training MLOps programs, including development of supporting REST APIs and services. Used Flask with Git and Docker as part of the delivery and deployment workflow.

REST APIs Flask Git Docker MLOps

Biomedical Engineer

AOV

Aug 2018 - Sep 2020 (2 years 1 month)

Worked on medical equipment maintenance and training MLOps programs, including building REST API components and related services. Used Flask with Git and Docker within the MLOps workflow.

REST APIs Flask Git Docker MLOps

Biomedical Engineer

Care Group

Aug 2018 - Sep 2020 (2 years 1 month)

Contributed to medical equipment maintenance and training MLOps programs, including REST API development and deployment workflows. Used Flask with Git and Docker as part of the MLOps delivery process.

REST APIs Flask Git Docker MLOps

Education

Degrees, certifications, and relevant coursework

University of Calcutta

Master of Technology (MTech), Autonomous Data Analyst

2016 - 2018

Activities and societies: LangGraph multi-agent text-to-SQL; governance routing; execution-repair; golden-set harness; RCA decomposition; OpenTelemetry cost tracing.

MTech at the University of Calcutta (2016–2018) building a LangGraph multi-agent text-to-SQL system with governance, validated SQL routing, execution-repair, and an eval-first harness. Included forced RCA decomposition, model tiering for cost control, and OpenTelemetry cost tracing.

WBUT

Bachelor of Technology (BTech), Multimodal Agentic Reasoning Assistant (MARA)

2011 - 2015

Activities and societies: Multi-agent multimodal assistant; hybrid retrieval (FAISS + BM25); adaptive planning with error recovery; FastAPI deployment.

BTech (2011–2015) from WBUT developing a multimodal agentic reasoning assistant coordinating RAG, vision, data, and web search agents for complex multi-modal queries. Implemented hybrid retrieval (FAISS + BM25), adaptive planning with error recovery, and a production-ready FastAPI service.