Hey, I'm Meghneel Gogoi

Software Engineer

Building scalable web applications, data pipelines, and applied AI/ML solutions.

Meghneel Gogoi profile pic

About

Final-year CCE student at Manipal University Jaipur (8.5 CGPA). Building end-to-end solutions across Full-Stack Development, Applied Machine Learning, and Generative AI.

Work Experience

Sypha logo

Sypha

Jan 2026May 2026

Software Development Intern

  • Led the v2.x → v3.x production update for Sypha AI Coding Extension, driving a SolidJS UI overhaul and legacy backend services integrating Keycloak PKCE authentication with silent token refresh and stale-while-revalidate profile polling.
  • Built a structured FastAPI analytics pipeline tracking per-message LLM costs and tool usage, and extended the extension with an SSE-backed fill-in-the-middle (FIM) autocomplete service.
  • Architected a Bun/Turborepo monorepo, extracting a multi-provider AI SDK with JWT-aware retries, concurrent refresh deduplication, and a Zod-validated model registry.
National Stock Exchange of India logo

National Stock Exchange of India

June 2025July 2025

Research Intern

  • Optimized a transcription pipeline using faster-whisper and CTranslate2, achieving 3x speedup on multi-lingual helpdesk audio processing.
  • Developed an in-house full-stack multilingual localization system using an NMT model with SQLite caching, reducing page translation time from 20s to milliseconds.

Featured Projects

University ERP System

University ERP System

Database-driven ERP platform for managing KPI workflows at MUJ.

TypeScriptNext.jsNest.jsReactTailwind CSSExpress.jsPostgreSQLPrismaSupabaseNextAuth.js

University ERP System

Designed and implemented a database schema and modular ERP layout tailored to university workflows at Manipal University Jaipur. Contributed to backend API development by building authentication routes and secure role-based access logic. Integrated departmental management and academic performance analytics into a scalable monorepo setup, leveraging modern frameworks for maintainability and performance.

AI Agent with RAG, Memory & Tools

AI Agent with RAG, Memory & Tools

AI agent with RAG, session memory, and plugin system built in a 1-day challenge.

TypeScriptNode.jsFastifyGoogle GeminiWeaviateZodMath.jsDocker

AI Agent with RAG, Memory & Tools

Developed a modular conversational AI agent in TypeScript with a Node.js + Fastify backend. Integrated Retrieval-Augmented Generation (RAG) using Weaviate for semantic search, Google Gemini for reasoning, and a persistent memory layer for contextual conversations. Designed a pluggable architecture supporting tools like weather and math plugins, enabling dynamic extensibility.

Customer Churn Prediction

Customer Churn Prediction

Boosted churn prediction accuracy to 81.65% by building an ensemble model combining Gradient Boosting, Logistic Regression, and AdaBoost.

PythonPandasScikit-learnNumPyMatplotlibSeaborn

Customer Churn Prediction

Developed a predictive analytics project targeting customer churn using an ensemble of Gradient Boosting, Logistic Regression, and AdaBoost. Achieved over 81% accuracy through rigorous preprocessing, feature engineering, and model tuning. Built a visualization pipeline to explain model outputs and improve stakeholder interpretability of retention strategies, highlighting actionable insights from data.

Web Translation Plugin

Web Translation Plugin

A server-side plugin based multi-lingual translation system for websites by extracting and translating content from the DOM.

JavaScriptTypeScriptExpress.jsSQLite3FastAPIDocker

Web Translation Plugin

Developed an in-house full-stack localization system capable of translating dynamic websites across 22 Indian languages. The system uses server side rendering for static pages and client-side script for dynamic pages to serve millions of users. A lightweight SQLite caching layer ensures sub-millisecond translations for repeated content, significantly improving and saving API calls to a self-hosted NMT model.

NDA

Research & Publications

1 publication
📄under-review
2025

A Hybrid Multi-Criterion Decision-Making Framework Using Intuitionistic Fuzzy AHP and VIKOR for Test Case Prioritization

Meghneel Gogoi, Aditya Narayan Hati, Dinesh Kumar Saini

International Journal of Information Technology & Decision Making (IJITDM)

This work proposes a hybrid decision-making framework for test case prioritization (TCP) that integrates Intuitionistic Fuzzy AHP (IF-AHP) and VIKOR. IF-AHP is employed to derive robust criterion weights by capturing uncertainty, hesitation, and subjectivity in expert judgments, while VIKOR is used to rank test cases by balancing conflicting criteria and identifying compromise solutions. The framework is validated on a multi-criteria software testing dataset, demonstrating its scalability, efficiency, and suitability for practical quality assurance environments..

Intuitionistic Fuzzy AHP (IF-AHP)VIKORTest Case PrioritizationDecision-Making FrameworkSoftware Testing

Technical Skills

Programming Languages

Python logo
Python
TypeScript logo
TypeScript
JavaScript logo
JavaScript
C++ logo
C++

Machine Learning & AI

PyTorch logo
PyTorch
Scikit-learn logo
Scikit-learn
LangChain logo
LangChain
LangGraph logo
LangGraph
LangFuse logo
LangFuse

Data Analytics & Visualization

Pandas logo
Pandas
NumPy logo
NumPy
Matplotlib logo
Matplotlib
Seaborn logo
Seaborn

Databases & ORM

PostgreSQL logo
PostgreSQL
MySQL logo
MySQL
SQLite3 logo
SQLite3
MongoDB logo
MongoDB
Prisma logo
Prisma

Vector Databases

Weaviate logo
Weaviate
ChromaDB logo
ChromaDB
FAISS logo
FAISS

Frontend

React logo
React
Next.js logo
Next.js
Tailwind CSS logo
Tailwind CSS
Vite logo
Vite
Shadcn/UI logo
Shadcn/UI

Backend

Node.js logo
Node.js
Express.js logo
Express.js
FastAPI logo
FastAPI

Cloud & Deployment

AWS logo
AWS
GCP logo
GCP
Vercel logo
Vercel
DigitalOcean logo
DigitalOcean

DevOps

GitHub Actions logo
GitHub Actions
Docker logo
Docker

VCS

Git logo
Git