// professional

Experience.

From a research-driven internship in semi-supervised gait analysis to my current role building data & AI products at AssetPLUS — here's what I've been up to.

  1. Current · loading…

    Data & AI Analyst

    AssetPLUS

    Active

    Building the data & AI layer at AssetPLUS — designing internal dashboards in Appsmith, automating cross-system workflows with n8n, and shipping LLM-powered features that turn raw data into business decisions.

    Internal Dashboards

    Designed and shipped Appsmith dashboards on top of MongoDB and SQL sources, giving teams real-time visibility into operations and KPIs.

    Workflow Automation

    Built n8n automations connecting databases, APIs, and notification channels — reducing manual effort and shrinking turnaround on recurring tasks.

    LLM Integration

    Integrated LLMs into product flows for summarization, classification, and assistive AI — with prompt design and evaluation against real data.

    Python & MongoDB

    Wrote production Python for ETL, analytics, and back-of-dashboard logic. Modeled MongoDB schemas tuned for the queries dashboards actually run.

    stack: Python Appsmith n8n MongoDB LLMs SQL JavaScript
  2. Internship · 2023 — 2024

    ML Research Intern — Predictive Modeling & Data Management for Gait Analysis

    M.Tech research project, VIT

    98% accuracy

    Led a semi-supervised gait-analysis project on human lower-limb movement — identifying gait cycles, segmenting them into seven phases, clustering with K-means, and training an RNN that hit 98% accuracy. Also built a Python Tkinter GUI on top of MongoDB to manage large datasets, with full CRUD and multi-format-to-CSV conversion for clean uploads.

    Gait Cycle Detection

    Used the slope formula on IMU sensor data to detect gait cycles in noisy real-world walking data.

    Phase Segmentation

    Applied a percentage-based method to split each gait cycle into seven distinct phases for downstream classification.

    Clustering & Labeling

    K-means clustering produced reliable phase labels (1–7), enabling supervised training on a previously unlabeled dataset.

    RNN Model

    Trained and evaluated an RNN classifier on the labeled phase data, achieving 98% accuracy on held-out samples.

    MongoDB + Tkinter GUI

    Designed a Python Tkinter desktop app on top of MongoDB with complete CRUD operations and multi-format file import.

    Data Pipeline

    Built a long-term data pipeline so future researchers can ingest, label, and analyze gait data without rebuilding tooling.

    stack: Python TensorFlow Scikit-learn MongoDB Tkinter NumPy Pandas
// open to chat

Looking for someone to build your data & AI layer?

I'm always happy to chat about dashboards, automations, and ML systems.

Get in touch LinkedIn