Lari

Portfolio · MMXXVI

MohdUmair
Lari.

AI/ML engineer & researcher engineering intelligent, cloud-native systems — from RAG pipelines and Monte Carlo finance to edge-native IoT security, published with IEEE and Springer.

View selected work→Get in touch
8.32
CGPA · B.Tech AI/ML
2
Research publications
300+
LeetCode submissions
3
Spoken languages
scroll ↓

01 — About

A builder who publishes.

I'm an undergraduate at Maulana Abdul Kalam Azad University of Technology, Kolkata, studying Computer Science with a specialization in Artificial Intelligence & Machine Learning.

My work sits at the intersection of applied ML, cloud-native engineering and research — shipping full-stack AI products on GCP and contributing to peer-reviewed papers on IoT security and malware detection.

Outside the terminal: I speak German (CEFR A2, sitting for B1), French (A1), and English fluently — and I'm always one coffee away from the next LeetCode contest.

02 — Toolkit

Stack & specialisms.

Languages

  • Python
  • Java
  • SQL
  • Bash
  • HTML / CSS / JS

AI / ML

  • scikit-learn
  • TensorFlow
  • LangChain
  • RAG Pipelines
  • Agentic AI
  • OpenCV

Google Cloud

  • Vertex AI
  • BigQuery / BigQuery ML
  • Cloud Run
  • Cloud Functions
  • VM Mapping
  • CI/CD

Engineering

  • FastAPI · Flask
  • MERN Full-Stack
  • REST APIs · CORS
  • MongoDB · MySQL
  • Docker
  • Git & GitHub

Concepts

  • SDLC
  • OOP
  • DSA
  • DBMS
  • Unit Testing
  • Version Control

Tools

  • Jupyter
  • MATLAB
  • Apache
  • Bootstrap

03 — Experience

Where I've worked.

  1. Apr 2024 — Jun 2024

    ICC & LOR awarded

    Data Analyst Intern

    Ozibook Fintech · Bengaluru

    • —Owned the core data-analysis project for Personal Coaches, designing solutions that increased client tenure and retention.
    • —Built Python tooling to validate large-scale financial datasets for 3 clients — lifted data accuracy by 38% and cut manual provisioning time across CI/CD.
    • —Engineered and maintained automated data pipelines in Python/Pandas, leading a team of 4 interns through development & deployment.
    PythonPandasCI/CDTeam lead

04 — Selected work

Projects.

01 / PROJECT

FinPass

AI-powered financial planning & investment optimization platform.

Full-stack AI application with multiple RESTful APIs (FastAPI + Flask) backed by MongoDB — personalised onboarding, portfolio analytics and investment guidance.

RAG-powered conversational responses with persistent memory, secure auth, and 1,000-run Monte Carlo goal simulations driving risk-based asset allocation.

CORS-aware cross-origin integration between a Vercel frontend and a HuggingFace-hosted backend.

FastAPIFlaskMongoDBRAGMonte CarloVercelHuggingFace
GitHub repo↗

02 / PROJECT

Serverless AI API on GCP

FinPass, productionised on Google Cloud.

Containerised Python backend + frontend with Docker, deployed to Cloud Run for fully managed, autoscaling serverless inference.

Integrated Google Generative AI via Vertex AI alongside pre-trained APIs, exposed through clean REST endpoints.

Leveraged BigQuery ML for data ingestion and analytics pipelines feeding the LLM layer.

Cloud RunVertex AIBigQuery MLDockerGenerative AI
GCP console↗

05 — Research

Published work.

  • IEEE MPCON 2026

    14–15 March 2026

    An Edge-Native Machine Learning Framework for Proactive IoT Network Security

    Babul P. Tewari, Umair Lari, Poulomi Mukherjee

    IEEE MPCON 2026

    Conference

  • Springer

    Accepted, 2026

    Malware Detection in IoT: A Machine Learning based Solution

    Umair Lari, Poulomi Mukherjee, Babul P. Tewari

    Springer — Securing the Internet of Things: ML and Beyond

    Book chapter

06 — Credentials

Certifications & achievements.

07 — Contact

Let's build something
worth publishing.

Email →Download CV (PDF)