Summary
Overview
Work History
Education
Skills
Websites
Personal Projects
Timeline
Generic
Mounish Sunkara

Mounish Sunkara

Chicago,US

Summary

Experienced AI Developer and Data Scientist with expertise in building enterprise-scale AI systems, machine learning models, and data analytics solutions. Strong background in LLMs, cloud technologies, and data engineering. Strategic professional versed in distilling and analyzing large data sets. Develops and delivers presentations detailing data findings. Articulate and collaborative with expertise in algorithm design and data collection.

Overview

7
7
years of professional experience

Work History

Senior AI Developer

Hendrickson USA L.L.C
Chicago, US
05.2023 - Current
  • Built an enterprise-grade, secure AI assistant (0 to 1) that effectively responds to user queries on organizational information, handling text, images, and tables, and reducing query resolution time.
  • Architected and deployed a scalable, multimodal document ingestion pipeline—processing text, images, and tables using Azure AI Document Intelligence, OCR, and transformers. Engineered with Durable Functions, and integrated Azure AI Search (vector DB) and Cohere embeddings for high-accuracy retrieval at scale.
  • Designed and deployed a cutting-edge Agentic RAG pipeline using advanced prompt engineering, LangGraph, and Azure OpenAI SDKs—delivered a highly optimized AI system with superior accuracy, speed, and reliability via FastAPI, and deployed through Docker and Azure Container Apps (ACA), and Kubernetes (AKS).
  • Implemented Server-Sent Events (SSE) to stream real-time status updates for the Agentic RAG process using a Redis Pub/Sub model, significantly reducing the perceived latency of the AI Assistant, and enhancing the user experience.
  • Built a robust, fault-tolerant LLMOps and eval framework using LangSmith—curated task-specific datasets, defined custom eval metrics, and conducted systematic experiments to rigorously benchmark and optimize RAG-based AI systems.
  • Enhanced image retrieval recall from 51% to 80%, driving a significant boost in user engagement with the AI Assistant.
  • Built an NL-to-SQL agent in LangGraph, providing senior executives with instant updates on sales orders and KPIs.
  • Implemented personalization (LLM enhanced) and privacy security measures, real-time encryption (AES-256) to secure user conversations, along with protections against prompt hijacking, jailbreaking, and other types of attacks, ensuring robust data privacy, and preventing unauthorized access to the AI system.
  • Generated account-level insights for sales teams in the truck parts manufacturing sector using LLMs, driving lead conversions, and integrating the solution into Salesforce and Power BI for streamlined data access and reporting.
  • POCs on AI agents, LangGraph agents frameworks, and extensive research on arXiv papers on AI systems.

Data Scientist

Philips
Bangalore, India
06.2018 - 06.2022
  • Built scalable machine learning models (forecasting) for the central warehouses to solve supply chain problems based on historical orders data. Used auto ARIMA models, ARIMAX models, and improved the fill rate by 26%.
  • Developed an anomaly detection model (from SAP data) to minimize defects in the manufacturing process using the Cost of Non-Quality (CONQ), Defects per Million Orders (DPMO), Statistical Process Control (SPC) methods such as control charts (X-bar chart and R chart, control limits), Pareto analysis, and Root-Cause analysis, and saved $480M in value.
  • Developed data models using PySpark in Delta Lake architecture and productionized Azure data pipelines—ETL pipelines (SQL Stored Procedures, ADF, and Azure Databricks)—and Power BI dashboards and Power Apps to track various KPIs, such as safety index, sustainability index, etc., for the maintenance of manufacturing plants.

Education

Master - Business Analytics

University of Cincinnati, Carl H Lindner College of Business
Cincinnati, US
05.2023

Skills

  • Python
  • Deep learning
  • AI system design
  • FastAPI
  • Generative Pre-training
  • Pandas
  • Numpy
  • PyTorch
  • LangGraph
  • Azure
  • AWS
  • Machine Learning
  • LLMs
  • PySpark
  • Power BI
  • ETL
  • Data Science
  • AI Development
  • Transformers
  • Natural language processing
  • Computer vision
  • Prompt engineering
  • Project management
  • Effective communication

Personal Projects

  • Implemented and trained a 124M parameter GPT-2 model from scratch in PyTorch on 10B tokens using distributed training across 8×A100 GPUs, achieving 72% loss reduction and 30.3% HellaSwag accuracy comparable to original GPT-2.
  • Built a character-level language model using Transformers architecture trained on Shakespeare's complete works to generate Shakespearean-style text, demonstrating proficiency in natural language processing and sequence modeling.
  • RAG Chatbot – Built a fully functional knowledge-base assistant using RAG and LangGraph; designed for rapid deployment and scalability in enterprise settings.
  • AI Comic Generation App Backend - convert a one liner concept to a comic with voiceover
  • Stock News Report Generation - customized stock news summary based on individual's exposure to the stocks.

Timeline

Senior AI Developer

Hendrickson USA L.L.C
05.2023 - Current

Data Scientist

Philips
06.2018 - 06.2022

Master - Business Analytics

University of Cincinnati, Carl H Lindner College of Business

Similar Profiles

CREATE PROFILE
Mounish Sunkara