✉ Open to opportunities
↗
Available for Opportunities
GET YOUR
AI ENGINEER
Maximizing development efficiency through an extensive ecosystem of modern AI tools. Specializing in LLM fine-tuning, multimodal GenAI, and autonomous AI orchestration.
Varun Likhitkar
AI Engineer
Specialization
AI/ML Specialist
Experience
Internship: Semester Tech
Education
B.Tech CSE — AIML
Python✦DeepSeek R1✦LLaMA✦Gemma 3.4✦n8n✦Flux✦ElevenLabs✦PyTorch✦Python✦DeepSeek R1✦LLaMA✦Gemma 3.4✦n8n✦Flux✦ElevenLabs✦PyTorch✦
Selected Work
PROJECTS
⌥
↗
Ovyis AI
Personal AI Platform
Engineered a hybrid personal AI platform deploying local LLMs for secure, private task management. Architected a local WLAN deployment model for seamless multi-device accessibility, and integrated global cloud infrastructure for remote access — balancing local data privacy with continuous availability.
View Project →
⌥
↗
AirTouch
AI Gesture-Controlled Virtual Mouse
Built an AI-based gesture control system using computer vision and MediaPipe for real-time mouse interaction. Added multi-hand tracking and motion smoothing for precise cursor control, drag-and-drop, and zoom actions. Packaged the Python application as a Windows executable for easy deployment.
View Project →
⌥
↗
Saha Traditions
Fashion E-Commerce Website
Built a fashion e-commerce website for Saha Traditions, a seller of salwar suits and sarees. Delivered a smooth shopping experience for browsing and showcasing ethnic wear collections, with a fully responsive client site to strengthen online presence and support sales.
View Project →
Technical Arsenal
SKILLS
LLMs &
Fine-Tuning
Fine-Tuning
- DeepSeek R1
- Gemma 3.4
- LLaMA 3.1
- LoRA / QLoRA
- PEFT
- Unsloth
Multimodal
GenAI
GenAI
- Flux (Image)
- Kling (Video)
- ElevenLabs TTS
- Whisper STT
- ComfyUI
- SDXL
AI
Orchestration
Orchestration
- n8n
- LangChain
- OpenAI Codex
- AutoGen
- CrewAI
- FastAPI
Core
Engineering
Engineering
- Python
- PyTorch
- OpenCV
- MediaPipe
- Git
- Docker
Work History
EXPERIENCE
AI Engineer Intern
Current Role
- Fine-tuned Gemma 3.4 and DeepSeek R1 using LoRA/QLoRA adapters on domain-specific datasets, achieving 23% improvement in task-specific accuracy.
- Built multimodal GenAI pipelines integrating Flux image generation and Kling video synthesis for content automation workflows.
- Designed and deployed n8n-based AI orchestration systems, reducing manual workflow time by 60%.
- Developed speech-to-speech AI assistant prototypes using Whisper ASR and ElevenLabs TTS with sub-500ms response latency.
Let's build something
GET IN
TOUCH.
TOUCH.