Data & AI Engineer

Preetham
Deekonda

M.S. Big Data Analytics. Building LangGraph agents, NLP microservices, and computer vision pipelines — for enterprise teams that ship.

View Work Resume ↗ Get in Touch
Overland Park, KS preethamdeekonda01@gmail.com +1 (913) 265-6451
$ python research_agent.py --query "RAG architecture"
✓ Router → query classified: retrieval
✓ Retriever → 42 chunks indexed
✓ Analyzer → cosine filter: 38 passed
✓ Synthesizer → answer + 3 citations
latency: 1.84s · source: local Ollama
$

About

I'm a Data & AI Engineer focused on moving past theoretical models into actual execution. My M.S. in Big Data Analytics gave me the foundation; my projects gave me the conviction that reliable, scalable AI systems are built from careful engineering — not hype.

I connect complex backend architecture with applied machine learning: whether that's a 4-node LangGraph agent graph, a TF-IDF categorization microservice, or a computer vision pipeline with real-time alerting. I build systems that solve immediate operational bottlenecks and scale reliably for enterprise teams.

4 Key projects
M.S. Big Data Analytics
Python Core language
Dec 2025 Graduated

Technical Skills

Languages
Python SQL Java JavaScript HTML/CSS YAML JSON XML
AI & Data
LangChain LangGraph RAG MCP A2A HuggingFace LLM Fine-tuning Pandas NumPy Scikit-learn OpenCV TF-IDF ETL Pipelines ChromaDB
Web & APIs
FastAPI Flask Spring Boot REST APIs Selenium Asyncio
Cloud & DevOps
Docker Kubernetes AWS S3 Lambda DynamoDB EKS MLflow Git Linux Wireshark

Work

01 · Jan 2026 – Mar 2026

Multi-Agent Research Assistant

Architected a 4-node LangGraph agent graph (router → retriever → analyzer → synthesizer) that classifies query intent, retrieves relevant document chunks, filters noise, and generates citation-backed answers. Built a RAG pipeline using ChromaDB for persistent vector storage and Ollama's nomic-embed-text for local embedding generation. Implemented a relevance scoring layer that ranks chunks by cosine similarity and filters below-threshold results before passing context to the LLM. Served through a FastAPI endpoint with streaming support, returning the answer, source metadata, and a step-by-step agent execution trace.

Python LangGraph ChromaDB Ollama FastAPI RAG
sub-2s retrieval across 10k+ document chunks
~35% reduction in hallucinated responses via cosine filtering
View project →
02 · Dec 2025 – Present

Autonomous Workflow Automation Engine

Built a high-concurrency Python engine maintaining 80% automation uptime on dynamic React frontends across multi-step tasks. Used NumPy physics vectors (gravity, friction) to simulate organic mouse movements, bypassing anti-bot detection. Applied Log-Normal statistical distributions to randomize click delays for long-running background tasks. Integrated TLS fingerprinting and strictly typed CSV serialization to ensure data integrity and pipeline security.

Python Nodriver NumPy Asyncio
80% automation uptime on dynamic React frontends
View project →
03 · Oct 2023 – Dec 2023

Content Recommendation & Clustering Engine

Automated news article categorization through a custom Python ETL pipeline and Scikit-learn clustering, accelerating the editorial publishing lifecycle by 30%. Built an ingestion layer consuming daily articles from third-party REST APIs, using Pandas for cleaning and normalization. Engineered a recommendation algorithm that maps user reading history to specific semantic content clusters. Visualized cluster distributions and engagement patterns in Power BI.

Python Pandas Scikit-learn Power BI REST APIs
30% acceleration in the editorial publishing lifecycle
View project →
04 · Feb 2022 – May 2022

Real-Time Intrusion Detection Pipeline

Built a computer vision pipeline in OpenCV using facial recognition algorithms to identify authorized personnel with real-time precision. Set up an automated SMTP event-response pipeline that instantly captures and transmits high-resolution evidence snapshots on anomaly detection. Created a lightweight Flask interface acting as a remote command center for toggling monitoring states. Integrated hardware and software to trigger real-time audio-visual alerts based on model confidence thresholds.

Python OpenCV Flask SMTP Real-Time
View walkthrough →

Experience

Machine Learning & Application Lead

Graduate Capstone · University of Central Missouri, Warrensburg, MO

June 2023 – Dec 2025
  • Reduced faculty administrative overhead by 40% by building a secure Java Spring Boot application that automated manual quiz grading, quiz generation, and question bank organization.
  • Developed a custom ETL pipeline using pdfplumber to extract text and tables from raw PDFs, enabling the system to ingest complex course materials without image noise.
  • Engineered a Flask microservice using TF-IDF vectorization to automatically map bulk-imported questions to textbook chapters, solving a primary categorization bottleneck.
  • Programmed an automated engine that reviews student quiz scores and generates personalized study reports pinpointing which chapters need review.
  • Secured the entire platform and all student data using custom Two-Factor Authentication (2FA) and Spring Security filters.

Education

M.S. Big Data Analytics & Information Technology

University of Central Missouri — Warrensburg, MO

Dec 2025

Contact

Let's talk.

Open to Data & AI engineering roles. I respond within 24 hours.

preethamdeekonda01@gmail.com +1 (913) 265-6451 Overland Park, KS