small lights,
patiently kept

SARATH BABU P — APPLIED MACHINE LEARNING, BENGALURU

explore

About

Behind the Data

I build things that hopefully work. Mostly ML systems, sometimes robots, occasionally chaos. Not the person who talks first in a room — but if you get me going about something I care about, good luck stopping me.

By day I'm writing Python that turns messy real-world data into something a team can actually trust. By night I'm either rewatching something I've already seen three times, on a volleyball court, or just existing quietly somewhere.

I got into this field by building things before I understood them. Still works.

“There is no secret ingredient. It’s just you.”

— Kung Fu Panda

“The things you own end up owning you”—

☕

Daily Fuel

Black. No exceptions.

🏐

Actual Sport

Volleyball (not just vibes)

🎬

Taste Level

Ice Age to Fight Club. All hits.

$ currently → building ML systems, one commit at a time

Download Resume

Projects

What I've Built

View All →

Build it.

Break it.

Fix it.

Ship it. →

Computer VisionDeep LearningPython

Cataract Detection System

Real-time cataract detection from eye images. Built the full pipeline — model, inference API, and serving layer. 98.1% accuracy.

98.1% detection accuracy on real eye image datasets

Computer VisionMobileOn-Device ML

Offline Facial Recognition System

100+ faces recognized entirely on-device. No cloud, no API calls, no internet required. Built with Flutter and embedding-based matching.

~95% accuracy, 100+ user embeddings, zero cloud dependency

Deep LearningAudio MLNLP

Audio Captioning System

Feed it audio. It tells you what's happening. CNN-LSTM and Transformer architectures on extracted audio features.

Contextual captions generated from raw audio without transcription

Experience

The Journey

Mooving (SAR Group)

R&D Engineer

Aug 2025 – Present

Built ML routing pipelines across a large fleet of field-service sites — 30% improvement in first-response efficiency using DBSCAN clustering and TSP route optimization
Automated recurring hardware-recovery audits, resolving stuck states without human intervention — ~25% improvement in uptime stability
Firmware analytics and anomaly detection at scale, running against real-world field data

Pythonscikit-learnFlaskMongoDBAPSchedulerPlotly

Livguard Drivetrain

Data Science Intern

Feb 2025 – Jun 2025

Processed 300M+ IoT logs and 1.6M+ transactions from field hardware deployed across 6 cities
Built automated anomaly-detection and reporting pipelines that ran without babysitting
Firmware benchmarking and degradation analysis — numbers ops teams actually used

PythonMongoDBPlotlypandasJupyter

Vmake

Embedded Systems Intern

Jan 2024 – Mar 2024

Built 12 IoT and robotics projects — gesture-controlled robots, LPG detectors, vitals monitors — generating ₹29K in revenue
Led a 6-person delivery team from prototype to handoff
/* [SARATH_FILL]: Add a third bullet if you have one */

ArduinoESP32Raspberry PiC++Embedded Systems