Charan Sai Ponnada
Building at the intersection of deep learning, computer vision, and production AI systems. IEEE published researcher crafting genomic foundation models and intelligent RAG systems.
// about
Building AI That Matters
I'm an AI Engineer and Machine Learning Researcher focused on deep learning, computer vision, and production AI systems. Currently building a genomic foundation model and working on LLM hallucination detection.
I am a B.Tech student in Artificial Intelligence & Data Science at VRSEC (Velagapudi Ramakrishna Siddhartha Engineering College) and an AI Engineer at Aynstyn Technologies, where I build production AI systems and SaaS platform features. My research spans computer vision, deep learning, and natural language processing — with publications at IEEE conferences.
I specialize in building RAG systems, fine-tuning vision-language models, and developing genomic foundation models using Mamba SSM architecture. I believe in creating AI that is not just intelligent, but reliable and trustworthy.
When I'm not training models or writing papers, I write technical articles on Medium, contribute to open source, and share my journey on YouTube at @charansimplifies.
Quick Facts
- Location
- Hyderabad, India
- Current Role
- AI Engineer @ Aynstyn
- Education
- B.Tech AI & DS, VRSEC
- Research
- IEEE Published
- Focus
- Deep Learning, CV, Genomic AI
- Writing
- Technical Articles & Tutorials
// skills
Tech Stack & Expertise
Technologies I work with daily and tools I reach for when building AI systems.
Programming Languages
ML & Deep Learning
Computer Vision
LLMs & NLP
Backend & APIs
Frontend
DevOps & Cloud
Databases
Tools & Frameworks
// experience
Timeline
My journey through AI engineering, research, and education.
2nd Place — YUVAAN 2026 Hackathon
IIT Hyderabad · Hyderabad, India · 2026-02 – Present
Secured 2nd place at IIT Hyderabad YUVAAN 2026 with VIVIRITY Intelli-Credit — AI-powered credit risk intelligence platform.
- Built ensemble ML system with 92% default prediction accuracy
- Implemented SHAP-based explainability for regulatory compliance
- Reduced false positives by 40% compared to traditional models
Published Researcher
ISAECT 2025 · International · 2025-11 – Present
Published research on Vision-Language Assistive Navigation using BLIP fine-tuning with 3-stage LoRA strategy.
- Fine-tuned BLIP model for visual-language assistive navigation
- Achieved +18% BLEU score gain with 3-stage LoRA
- Published and presented at ISAECT 2025
AI Engineer
Aynstyn Technologies · Hyderabad, India · 2025-08 – Present
Building production AI systems and SaaS platform features. Implementing PPO-secured customer handling pipelines and AI-powered automation.
- Developed and deployed PPO-secured AI customer handling pipeline
- Built production-grade SaaS features serving 10+ enterprise clients
- Implemented real-time data processing with FastAPI and Redis
- Containerized microservices with Docker on AWS ECS
B.Tech in Artificial Intelligence & Data Science
Velagapudi Ramakrishna Siddhartha Engineering College (VRSEC) · Vijayawada, India · 2023 – 2027
Pursuing Bachelor of Technology in Artificial Intelligence and Data Science with focus on Deep Learning, Computer Vision, and NLP.
- Specialization in AI and Data Science
- Published IEEE research papers at ISAECT 2025
- Active in open-source contributions and technical writing
- Built production AI systems at Aynstyn Technologies
// research
Published Research
Exploring the frontiers of AI through peer-reviewed research. IEEE publications in computer vision, NLP, and deep learning.
Vision-Language Assistive Navigation for Visually Impaired Using BLIP Fine-Tuning
Fine-tuned BLIP model using 3-stage LoRA for visual-language assistive navigation. Published at ISAECT 2025.
BLEU Score Gain
+18%
Dataset Size
50K pairs
LoRA Stages
3-Stage
Inference Speed
2.5x
Semantic Consistency for Hallucination Detection in Large Language Models
Novel framework for detecting hallucinations in LLMs using semantic consistency checking. Under review at IEEE InCODE-2026.
Detection F1
0.89
Models Evaluated
5 LLMs
Benchmarks
3 Datasets
Improvement
+12% F1
// featured work
Projects & Systems
Selected projects that showcase my work across AI, ML, research, and full-stack development.
VIVIRITY Intelli-Credit
AI-powered credit risk intelligence platform. Built for YUVAAN 2026 at IIT Hyderabad — secured 2nd place.
AyurMind
Domain-specific RAG system for Ayurveda. Retrieval-augmented generation over ancient medical texts.
Aynstyn Technologies Platform
Production SaaS platform for AI-powered business solutions. Built during AI Engineer internship.
Genomic Foundation Model
In ProgressMulti-species genomic foundation model using Mamba SSM architecture. ~100M parameters.
// writing
Latest Articles
Thoughts on AI, machine learning, deep learning, and building production systems.
Understanding Mamba SSM: The Architecture That Could Replace Transformers
A deep dive into state-space models, the Mamba architecture, selective scan algorithm, and why it matters for efficient sequence modeling.
Building Production RAG Systems: A Practical Guide
Everything you need to know about building retrieval-augmented generation systems that actually work in production.
Fine-Tuning Vision-Language Models: A 3-Stage LoRA Approach
Learn how to efficiently fine-tune BLIP and other vision-language models using parameter-efficient techniques.
// medium
Latest on Medium
Articles I've published on Medium about AI, machine learning, and data science.
Loop Engineering? Lets Clear the Things With This
We spend hours prompting AI systems. Prompt. Review. Fix. Prompt again. What if we could automate that iteration itself?
My Experience Fine-Tuning a Vision-Language Model on Low Compute
Fine-tuning a VLM sounds exciting — until you try doing it with limited data and limited compute. Lessons from the trenches.
// faq
Frequently Asked Questions
Quick answers to common questions about my work, research, and collaborations.
My research focuses on computer vision, deep learning, and natural language processing. I work on vision-language models, hallucination detection in LLMs, and genomic foundation models using state-space architectures.
Yes! I am always open to research collaborations, project partnerships, and consulting opportunities. Reach out via email or LinkedIn.
Python, PyTorch, TensorFlow, LangChain, FastAPI, React/Next.js, Docker, AWS, Mamba SSM, Hugging Face, ChromaDB, and more. See my skills section for the full list.
Yes, I have 1 published IEEE conference paper and 1 paper currently under review. My research covers vision-language assistive navigation and LLM hallucination detection.
A genomic foundation model is a large-scale AI model trained on DNA sequences from multiple species. It learns evolutionary patterns and can be fine-tuned for downstream tasks like variant effect prediction, gene regulation understanding, and species classification.
Yes, I write technical articles on AI, ML, and deep learning. I also run a YouTube channel (@charansimplifies) where I share tutorials and insights about AI engineering.
// contact
Let's Build Something
I'm always open to discussing research collaborations, project ideas, or opportunities in AI and ML.
Whether you have a research collaboration in mind, a project you want to build, or just want to say hi — I'd love to hear from you.
charansaiponnada06@gmail.comBased in Hyderabad, India. Available for remote opportunities worldwide.