Anugrah Ajith
000
CS Engineer · Kerala, India

Anugrah
Ajith.

Final-year CS student at LPU. AI Evaluation Specialist at Scaler AI Labs. Building systems that think, interfaces that resonate, and pipelines that scale. Harmonizing technical complexity with intuitive design.

3+
Years Building
420k
Records Modelled
0.95
R² Score
LLM Alignment RLHF Training XGBoost / SHAP FastAPI FFmpeg TensorFlow AI Evaluation Adaptive Streaming Red-Teaming Research Author System Design LLM Alignment RLHF Training XGBoost / SHAP FastAPI FFmpeg TensorFlow AI Evaluation Adaptive Streaming Red-Teaming Research Author System Design
01

Professional Trajectory

Dec 2025 — Present
AI Evaluation Specialist
Scaler AI Labs

Developing SFT "Golden Responses" for LLM reasoning benchmarks and architecting automated annotation pipelines that ensure frontier-model alignment at scale.

Current
Mar 2026 — Present
Business Development
Acmegrade

Bridging technical product capabilities with market-driven user needs — translating complex AI features into compelling value propositions.

Current
Feb 2025 — Jul 2025
Senior AI Trainer
Outlier (Scale AI)

Aligning frontier LLMs via RLHF and performing adversarial red-teaming for algorithmic reasoning — shaping how the next generation of AI systems behave.

Completed
Jan 2025
Research Author
Intl. OS Research Journals

Published "Novel Users in Specific Domains" — an exploration of specialized user behaviour patterns in OS environments, bridging HCI theory with practical systems design.

Published
02

Technical Builds

Project / 001

GeoAI: Air Quality Forecasting

XGBoost · SHAP · Python

Predictive air-quality model trained on 420,000+ geospatial records with full explainability via SHAP values. Production API with real-time inference.

0.9579 R² Score
Project / 002

Distributed HLS Streaming

FastAPI · FFmpeg · HLS

Adaptive bitrate streaming service supporting high-density video delivery and real-time transcoding workflows. Handles concurrent session scaling gracefully.

ABR Adaptive Bitrate Engine
Project / 003

ML Behavioural Security

Scikit-learn · TensorFlow

Anomaly detection prototype that identifies browsing-based threats and zero-day attack patterns using behavioural signals rather than signature matching.

0-day Pattern Detection
03

Philosophy & Stack

"I don't build features — I build systems that earn trust and interfaces that feel inevitable."

I'm a final-year Computer Science student at Lovely Professional University, currently working as an AI Evaluation Specialist at Scaler AI Labs — where I shape how frontier language models reason, align, and behave.

My work sits at the intersection of machine learning, systems engineering, and product intuition. Whether it's crafting SFT golden responses, building XAI pipelines, or architecting adaptive media systems — I'm drawn to problems where elegance and rigor have to coexist.

I believe the best engineers are also designers — people who obsess over how something feels as much as how it works.

Python XGBoost TensorFlow FastAPI RLHF SHAP / XAI FFmpeg Scikit-learn LLM Alignment Red-Teaming HLS Streaming System Design