Chebrolu Trilokesh Venkata Uday — Big Data & AI Engineer

01 Profile

About

Data Science-focused B.Tech Computer Science student and Associate Software Engineer with hands-on experience designing and shipping enterprise-scale AI analytics platforms, multi-agent LLM systems, and big data pipelines. Proficient across the full stack of modern data engineering and applied AI: natural-language-to-SQL systems, RAG pipelines, semantic layers, and MLOps frameworks, using tools including Apache Spark, Snowflake, BigQuery, Kafka, Docker, and LangGraph/CrewAI. Published researcher with a co-authored book chapter on web application security. Experienced in taking products from architecture through production deployment for real-world users at scale.

NL-to-SQL RAG Pipelines Semantic Layers MLOps Multi-Agent Systems Data Lakehouse

02 Professional Experience

Where I've worked

Associate Software Engineer

Feb 2026 – Present

Euron

VoteEdge — AI-Powered Election Campaign Management Platform (Nigeria)

Built a multi-tenant SaaS platform covering 93.4M+ registered voters, 176,974 polling units, 8,809 wards, and 774 LGAs across Nigeria.
Architected a microservices system: React/Next.js web app, offline-first React Native mobile app (WatermelonDB, CRDTs, 72-hour offline support), Node.js/FastAPI backend, and Kong/AWS API Gateway.
Implemented multi-LLM routing (LiteLLM) across Claude, GPT-4o, and Gemini for OCR-based result collation, predictive voter scoring, sentiment analysis, and voice/chat agents, avoiding the need for custom ML model training.
Built an event-driven data pipeline using Kafka feeding a Spark + Delta Lake data lakehouse, with Elasticsearch for search and Redis for caching and LLM cost reduction via semantic caching.
Deployed a multi-region AWS architecture (Lagos primary, Ireland DR) with PostgreSQL + PostGIS for geospatial queries; delivered an 8-module platform in a compressed 4.5-month timeline.

GURU — AI Teaching Operating System

Designed a multi-agent AI architecture spanning 7 engines and ~19 agents, coordinated through a central Student Intelligence Graph, for personalized K-12 and competitive-exam education.
Built CARA, a voice-first conversational AI tutor supporting 22+ Indian languages and dialects with Socratic-method tutoring and real-time misconception detection.
Built SOMA, a real-time student modeling engine that tracks per-concept mastery (0–100), cognitive load, and learning-style (VARK) via a live knowledge-state graph.
Built EVAL, an assessment engine with confidence-weighted scoring, mastery gates, and predictive board-exam performance forecasting, and MEMO, a spaced-repetition engine (SM-2/FSRS) with automatic flashcard generation.
Built BRIDGE, a communication engine generating automated parent/teacher reports in vernacular languages with at-risk student alerts and DigiLocker/UDISE integration, targeting a 260M+ student addressable market.

Model Verse — B2C AI Marketing Asset Generation Platform

Built a brand-to-asset generation platform (Next.js 14, Tailwind, Shadcn/ui, Zustand + React Query) that generates marketing assets directly from a brand URL.
Designed a microservices architecture (API gateway plus User, Project, Brand, Asset, and Billing services) with async generation workers backed by PostgreSQL, Redis, and S3-compatible storage.
Integrated multiple AI providers with fallback routing — Stability AI/Replicate for image generation, Replicate IDM-VTON for virtual try-on, Runway ML/Pika Labs for video, and GPT-4 Vision/Claude for brand and competitor analysis.
Delivered a ~90% cost reduction versus traditional photoshoots and cut asset production time from weeks to minutes.

Big Data Engineer Intern

Nov 2025 – Feb 2026

Euron

Self-Service Analytics Platform (Google BigQuery)

Built a multi-tenant, AI-driven self-service analytics platform combining RAG-based document Q&A with NL-to-SQL generation over BigQuery, accessible via a React frontend and Slack bot.
Implemented vector search using PostgreSQL pgvector and Pinecone for semantic document retrieval, with Google Drive and SharePoint as connected document sources.
Built full tenant isolation with RBAC and enterprise SSO (Okta, Azure AD, Google Workspace), plus SQL transparency features including syntax-highlighted queries, execution timers, confidence scoring, and source citations.
Delivered the platform on a 24-week timeline with monitored system health across the RAG pipeline (1.2s avg), SQL generator, and BigQuery layer (850ms avg).

Snowflake Agentic Analytics Pipeline

Designed a multi-agent pipeline (LangGraph/CrewAI) handling intent routing, schema exploration, SQL generation, validation, execution, and insight/dashboard generation, with governed writeback.
Built a self-correcting NL-to-Snowflake-SQL agent that validates queries in a LIMIT-100 mode and re-prompts on error, backed by a semantic layer (facts, dimensions, KPIs) using OpenAI embeddings for schema mapping.
Implemented a governed writeback flow (validate → submit → approve → merge) generating INSERT/MERGE statements against whitelisted Snowflake views, enforcing row-access and data-masking policies with full audit logging.
Built a React 18 + TypeScript frontend with a drag-and-drop dashboard designer (ECharts/Recharts, AG Grid) and GPT-4.1-powered narrative insight generation.

03 Personal Projects

What I've built on my own

Alpha-Z — Automated Trading Strategy Execution Platform

Containerized strategy execution using Docker, ensuring isolation and dependency consistency, with automated execution lifecycle management and resource-optimized auto-termination.
Designed secure broker API credential management compliant with new SEBI regulations, supporting dynamic token refresh; deployed on AWS with extensibility for multi-broker support.

DockerAWSPythonBroker APIs

VinoFlow — End-to-End MLOps Pipeline for Wine Quality Prediction

Designed and deployed a robust MLOps pipeline ensuring automation and reproducibility across the ML lifecycle, implementing Jenkins CI/CD for continuous training, testing, and deployment.
Used DVC for dataset and model versioning, tracked experiments with MLflow, containerized services with Docker, and exposed the final model as a scalable Flask REST API.

JenkinsDVCMLflowDockerFlask

Automated Sentiment Analysis Pipeline with CI/CD

Developed a full-stack application for real-time sentiment analysis, operationalizing the ML lifecycle from experimentation to automated deployment, with MLflow for experiment tracking and NLP library comparison.
Built a CI/CD pipeline using GitHub Actions to automate linting, testing, and deployment; containerized the system with Docker and deployed an interactive Streamlit dashboard.

MLflowGitHub ActionsDockerStreamlit

04 Education

Academic background

Bennett University

Bachelor of Technology, Computer Science and Technology

Greater Noida, India
09/2022 – 06/2026
CGPA: 8.9/10

05 Technical Skills

Tech stack

Py

Python

C++

SQL

DSA

Data Structures & Algorithms

PM

Predictive Modeling

NLP

DL

Deep Learning (LSTM, CNN)

TF

TensorFlow

PT

PyTorch

SK

Scikit-learn

LG

LangGraph

CA

CrewAI

SP

Apache Spark (PySpark)

Hd

Hadoop

Hv

Hive

Ka

Kafka

Db

Databricks

Sf

Snowflake

BQ

BigQuery

My

MySQL

Mo

MongoDB

PG

PostgreSQL (pgvector)

Pr

Presto

DW

Data Warehousing

Do

Docker

Je

Jenkins

CI

CI/CD Pipelines

ML

MLflow

Af

Airflow

DVC

Git

GA

GitHub Actions

Ta

Tableau

St

Streamlit

Mp

Matplotlib

Sb

Seaborn

06 Publications

Research

Web Application Security Assessment Techniques: Current Trends and Future Directions

Book Chapter · Intelligent Computing and Communication Techniques · CRC Press (Taylor & Francis Group), 2025

Co-authored a book chapter providing an in-depth review of state-of-the-art web application security assessment approaches, covering static and dynamic analysis, penetration testing, automated scanning, and integrated methodologies, and discussing emerging practices and future research directions.