Arjungopal Anilkumar — AI/ML Engineer

AI / ML Engineer

Arjungopal
Anilkumar.

LLMs · Deep Learning · Cloud Deployment

I build production-grade AI systems — from fine-tuned language models and multi-agent pipelines to cloud-deployed diagnostic tools that deliver measurable outcomes.

Currently at Amrita Vishwa Vidyapeetham, I specialise in RAG architectures and deploying deep learning on cloud infrastructure. Every project prioritises production readiness: secure pipelines, quantifiable accuracy improvements, and real-world data at scale.

Download Resume Get in Touch View Projects ↓

Skills.

Core technologies I use to design, train, and deploy intelligent systems in production.

LLMs & GenAI

Prompt Engineering LangChain Hugging Face OpenAI API Ollama LoRA / PEFT Fine-Tuning RAG Pipelines Transformers Multi-Agent Systems

Deep Learning & Data

TensorFlow Keras Scikit-learn Machine Learning CNNs NER Pandas NumPy EDA Feature Engineering

Infrastructure & Data Engineering

Python SQL / PostgreSQL Apache Kafka Apache Spark Airflow HDFS Grafana Docker Azure FastAPI REST APIs Git Java C

Education.

Academic foundation in AI-specialised computer science, with strong performance across core and advanced coursework.

2023 – Present

Amrita Vishwa Vidyapeetham

B.Tech — Computer Science Engineering (AI)

Coimbatore, Tamil Nadu · CGPA 7.72

Deep Learning Natural Language Processing Data Structures & Algorithms Cloud Computing Database Management Computer Vision Operating Systems Software Engineering

2021 – 2023

St. Jude Public School

Higher Secondary — Computer Science Stream

Thrissur, Kerala · 86%

Experience.

Professional roles and leadership positions where I delivered real impact.

Work Experience

Paramount Computer Systems

Artificial Intelligence Engineer Internship Apr 2026 – Present Hybrid

Engineered and debugged core AI module pipelines within an Agile SCRUM environment, executing rigorous end-to-end UAT testing to minimize deployment bug frequency.
Architected and documented scalable codebase components for internal enterprise systems, optimizing API query response times and overall system latency.

Cycrew

AI Engineer — Security Intelligence Internship Oct 2025 – Jan 2026 Remote

Fine-tuned LLMs (Ollama, Hugging Face) with LangChain orchestration and OpenAI APIs — achieved a 35% improvement in threat-classification accuracy.
Built automated log-analysis pipelines processing 5,000+ security events, reducing mean triage time by 25%.
Developed RAG-powered summarisation workflows for phishing campaigns, MFA abuse patterns, and threat intelligence briefs.

Leadership & Extracurricular

The ELITE Club

Head of Event Management Dec 2025 – Present Coimbatore

Led end-to-end execution of 5+ technical and cultural events with 500+ cumulative attendees; streamlined cross-team coordination workflows.

Youth United Council of India

Campus Head Mar 2025 – Jun 2025 Coimbatore

Directed a 10-member committee; planned and executed campus-wide events reaching 300+ participants within a single quarter.

The Institution of Electronics and Telecommunication Engineers

Event Management Team Member Aug 2025 – Present Coimbatore

Designed and ran structured onboarding for 100+ incoming members, increasing first-event participation rates by 40%.

Youth United Council of India

Event Management Team Member Dec 2024 – Mar 2025 Coimbatore

Owned logistics and scheduling for 200+ attendee events; established operational playbooks adopted by subsequent teams.

Publications.

Research on NLP and computational linguistics, focusing on linguistically-informed architectures.

LIMP: Linguistically-Informed Multi-Strategy Prompting for Telugu Multi-Turn Dialogue Generation

Arjungopal Anilkumar, Suryansh Ram Menon

ACL 2026 · Proceedings of the Sixth Workshop on Speech, Vision, and Language Technologies for Dravidian Languages · July 2026

Generating contextually coherent multi-turn dialogue in Telugu requires resolving three deeply interacting constraints absent from generic LLM prompting: morphologically encoded social hierarchy, strict SOV agglutinative syntax, and culturally governed emotional logic formalised in Natyashastra rasa theory. We introduce LIMP, an inference-time, training-free framework that injects expert linguistic and cultural knowledge into prompt structure. We empirically evaluate two strategies on 10,000 stratified evaluation instances, demonstrating that sequential analytical commitment to linguistic constraints produces significantly more form-faithful Telugu than holistic constraint injection.

PDF Paper Google Scholar