Software Engineer · Data Engineering · Cloud · AI/LLM Apps

Results-driven Software Engineer with 4+ years building scalable data products and full‑stack solutions across Snowflake, DBT, Airflow, AWS, and Spark—plus hands-on AI/LLM work (RAG, local models, LangGraph). I enjoy turning ambiguous data and platform problems into reliable, production-grade pipelines and developer-friendly apps.

• Current focus: DBT/Snowflake data products, Airflow orchestration, EMR ops, cost‑efficient LLM agents.
• Strengths: data modeling, pipeline performance, workflow reliability/SLA hygiene, pragmatic AI integrations.
• Open to: data engineering roles, platform/infra data work, privacy-first AI applications.

• Data/Cloud: Snowflake, DBT (Core & Cloud), Airflow (MWAA), Databricks, Hadoop, Spark, Hive, AWS (S3, EMR)
• Programming: Python, SQL, Shell, JavaScript/TypeScript, Java, C/C++
• Web: React, Node.js, Next.js, FastAPI, Laravel, WordPress
• Databases: Oracle, PostgreSQL, MySQL, MongoDB, ChromaDB
• AI/LLM: LangGraph, Streamlit, RAG, Vector DBs, Ollama, LLaMA/Mistral

• Built and maintained DBT+Snowflake data products for enterprise analytics with robust testing, governance, and performance tuning.
• Operated Airflow (MWAA) and Arena environments with incident response, EMR reliability, and end‑to‑end workflow SLAs.
• Migrated DBT Core → DBT Cloud with Airflow DAG refactors, validation, and observability.
• Delivered privacy-first AI apps: local LLM RAG and a hybrid token‑efficient agent.
• Full‑stack background: React/Next.js + FastAPI/Laravel, strong API design and docs.

1. DocuRAG: Document Analysis Platform
   React + FastAPI + PostgreSQL + ChromaDB + Ollama (LLaMA 3.2)

• Local, privacy-first RAG for PDFs/TXT/DOCX with vector embeddings and on‑device inference.
• Dockerized, Nginx‑backed deployment; supports multi-user auth and knowledge bases.
• Useful for research analysis and internal KB search.

2. TokenSage: Hybrid LLM Agent
   LangGraph + Streamlit + OpenAI API + Local LLMs (Ollama)

• Minimizes premium token usage by delegating expansion to local models.
• Orchestrates API+local responses for quality and cost control.
• Applied to docs, support, and education content generation.

3. DataGene: Synthetic Data Generator
   Java + Spring Boot + RabbitMQ + S3

• High‑volume data synthesis with secure S3 storage and rich observability.
• Admin dashboard for environment variables and connections.

4. RAG Pipeline for Jira Ticket Insights
   LangChain + Streamlit + LLaMA + ChromaDB

• Ingests Jira ticket dumps, indexes features, and predicts likely error occurrences.
• Fast, searchable interface for triage and analysis.

• Alexion Operations – Data Products (Snowflake, DBT, MWAA/Airflow): DBT model design, testing, performance; data governance artifacts (Collibra), conceptual models; end‑to‑end solution testing and production handover.
• BofA Product Support (Arena): Incident handling (JIRA), log analysis, Spark upgrade support, platform stability.
• DBT Core → Cloud migration: Refactored Airflow DAGs to DBT Cloud jobs, validated lineage and outputs.
• Managed Services (Arena, Airflow, EMR, AWS): Proactive monitoring, reruns, biweekly maintenance, infra/app issue ownership.

• Software Engineer, TrieDatum — Apr 2023 → Present
• Software Engineer, Ekodus Technologies — Jul 2021 → Apr 2023
• Software Engineer, Riverrun Digital — Sep 2020 → Jun 2021

Education: M.Sc. Information Technology, Gauhati University
Certifications: Snowflake (Udemy, 2023), DBT (Udemy, 2024)

• Email: [email protected]
• LinkedIn: linkedin.com/in/alokdeka