Kiren S

AI Engineer • LLM Systems • Production ML

Building production-grade AI systems — from LoRA fine-tuning LLMs to deploying real-time NLP pipelines serving live users.

KS

About Me

Final-year B.E. AI & Data Science student (CGPA: 8.32) with production experience across the full AI engineering stack — LoRA fine-tuning LLMs, building RAG pipelines, deploying FastAPI inference APIs, and shipping full-stack AI applications to 100+ concurrent users.

Currently building a real-time finance news sentiment system using FinBERT, targeting Indian fintech startups. Seeking an AI Engineer role to build and ship production-grade AI systems.

4+
Projects Deployed
8.32
CGPA
100+
Concurrent Users Served
5
Certifications
KS

Technical Skills

Python

LLM Fine-tuning (LoRA)

Hugging Face / FinBERT

LangChain / RAG

FastAPI

Docker

PyTorch

scikit-learn / FAISS

OpenCV (C++ & Python)

SQL / SQLite

Pandas / NumPy

Git

Selected Projects

RAG Chatbot using LangChain

End-to-end Retrieval-Augmented Generation pipeline — PDF ingestion, chunking, HuggingFace sentence-transformer embeddings, FAISS vector storage, and context-grounded LLM response generation. Supports multi-turn document Q&A with hallucination-reduced answers.

LangChain FAISS Hugging Face FastAPI

Number Plate Detection & OCR

Two-stage ANPR pipeline: YOLOv8 for Indian license plate localization followed by fine-tuned PaddleOCR for text extraction, exported to structured CSV for downstream processing.

YOLOv8 PaddleOCR Python

Edge Package Dimension Estimator

Classical CV pipeline in C++ with OpenCV: HSV masking → Gaussian blur → Otsu thresholding → morphological closing → contour extraction → bounding box estimation with real-world cm calibration. Production-grade single-camera measurement system.

C++ OpenCV Computer Vision

Work Experience

Oct 2025 – Present

UIT Global Solution

AI Engineer Intern

  • Fine-tuned LLaMA 8B Instruct using LoRA + Unsloth, reducing training time ~40% while achieving strong domain-specific instruction-following.
  • Designed and deployed full-stack AI application (NestJS + React), load-tested to 100+ concurrent users with low-latency inference endpoints.
  • Built and maintained production ML inference pipelines — model versioning, request handling, and performance monitoring.
2025

App Innovation Technologies

Data Analytics Intern

  • Built multi-class text classification models using BERT (HuggingFace) and scikit-learn pipelines with systematic hyperparameter optimization.
  • Developed interactive dashboards using Matplotlib and Seaborn to surface actionable trends for non-technical stakeholders.

Education & Certifications

2026

United Institute of Technology, Coimbatore

B.E. in Artificial Intelligence & Data Science

CGPA: 8.32
2024

Udemy

Machine Learning A-Zā„¢

Comprehensive ML techniques and applications.

2024

Udemy

Deep Learning Specialization

Neural networks, CNNs, and deep learning frameworks.

2024

Udemy

Python for Data Science & ML Bootcamp

Data analysis, visualization, and applied ML.

2024

Accenture & Tata Group (Forage)

Data Analytics Virtual Internships

Data-driven insights and business optimization projects.

Let's Connect

Message sent successfully!