ATHARVA RAHATE

AI Engineer | LLM Optimization, Agentic AI & Computer Vision

Applied AI Research • Production ML Systems • System Design • Technical Leadership

        

---

PROFESSIONAL SUMMARY

SPECIALIZATION Model Optimization & Inference Acceleration Computer Vision Pipeline Development Retrieval-Augmented Intelligence Systems Production ML System Architecture

IMPACT METRICS 98% Accuracy Vision Models 42% Inference Time Reduction 95%+ Production Model Accuracy Top 200 of 68,000 in Analytics

AI Engineer focused on building efficient, production-ready systems across LLMs, agentic AI, computer vision, and applied ML. My work centers on model optimization, retrieval-augmented intelligence, and designing pipelines that run reliably at scale. Currently exploring agentic workflows, retrieval systems, and large-scale operational networks to design context-aware AI solutions that work beyond controlled environments.

---

TECHNICAL EXPERTISE

Core Technologies

AI/ML Frameworks

LLM & Agentic AI

Development Stack

DevOps & Data

---

PROFESSIONAL EXPERIENCE

Devspace IT Solutions AI Engineer Oct 2025 - Present	Developed YOLO-based crop disease detection system achieving 98%+ accuracy across 5 crop families Scaled detection pipeline from single dataset to 5+ crop varieties without precision loss Enhanced real-world inference reliability through advanced preprocessing and noise-handling mechanisms Enabled faster agronomic diagnostics for production agriculture applications
Tactiqe.in Software Engineer Oct 2025 - Present	Architected student-led engineering community platform with enterprise-level workflows Designed and deployed scalable platform architecture enabling structured project collaboration Established CI/CD pipelines, PR guidelines, and engineering standards improving code quality Mentored cross-functional teams (AI, Backend, Frontend) on production-ready development practices
Celebal Technologies Data Science Intern May 2025 - Jul 2025	Optimized transformer-based LLMs with TensorFlow, reducing inference time by 42% and compute usage by 18% Built ensemble and Random Forest models achieving 95%+ accuracy on production datasets Automated deployment for two production ML systems using Docker and CI/CD pipelines Enhanced data pipelines through improved preprocessing, feature engineering, and quality checks
Internship Studio ML Engineer Sep 2024 - Nov 2024	Built YouTube Content Analyzer using CNNs for multi-class video content detection Optimized preprocessing and feature extraction pipeline for diverse video quality conditions Integrated object-detection components for real-time and near-real-time analysis Strengthened ML engineering fundamentals through dataset preparation and iterative model tuning

---

RESEARCH & DEVELOPMENT PROJECTS

INTERVEE - AI Mock Interview Simulator

Technical Stack: Python, NLP, Speech Recognition, Deep Learning, FastAPI, React

Advanced NLP and voice analysis system improving candidate feedback accuracy by 50%

Key Features:

• Real-time speech-to-text processing with sentiment analysis
• Multi-modal evaluation framework (verbal content, tone, delivery patterns)
• Actionable performance insights dashboard with scoring breakdowns
• Question generation engine using transformer models
• Voice emotion detection using audio feature extraction

Impact: Enhanced interview preparation experience with data-driven feedback

HEALTHPREDAI - Medical Risk Prediction System

Technical Stack: Python, Scikit-learn, XGBoost, SHAP, Flask, SQLite

ML-based risk assessment integrating tabular, text, and rule-based signals

Key Features:

• Ensemble models with interpretable SHAP explanations
• Multi-source data fusion for comprehensive health risk scoring
• Privacy-first architecture with federated learning concepts
• Clinical decision support system integration
• Real-time risk score computation with confidence intervals

Impact: Provides actionable health insights with transparent reasoning

STEGA-SHIELD - Secure Steganography System

Technical Stack: Python, Cryptography, PIL, Flask, React, WebSockets

Full-stack encrypted data hiding and secure media transmission platform

Key Features:

• AES-256 encryption for embedded data
• Support for images, audio, and video media formats
• LSB (Least Significant Bit) embedding algorithm
• Web-based interface with real-time encoding/decoding
• Secure key exchange mechanism

Impact: Enables secure covert communication with military-grade encryption

LUNAR CRATER DETECTION - Deep Learning Research

Technical Stack: Python, TensorFlow, OpenCV, Satellite Imagery Processing

Geospatial crater mapping using custom CNN architecture for Avishkar Research Competition

Key Features:

• Transfer learning from ImageNet weights with custom head
• Data augmentation pipeline for limited lunar imagery datasets
• Precision crater boundary detection with instance segmentation
• Geospatial coordinate mapping integration
• Model achieved finalist recognition at state-level competition

Impact: Improved crater mapping precision for lunar surface analysis

GENERATIVE ARCHITECTURE DESIGN - Patent Filed

Technical Stack: Python, GANs, Transformers, 3D Modeling Libraries

Automated design reasoning with multimodal generation capabilities

Key Features:

• Text-to-3D model generation pipeline using diffusion models
• Architectural constraint satisfaction solver
• Integration with CAD export formats (OBJ, STL, DXF)
• Style transfer for architectural aesthetics
• Parametric design optimization

Impact: Patent filed for automated architectural design generation system

---

ACHIEVEMENTS & RECOGNITION

Competition Performance Event Rank Participants National Hackathon (Blockchain) 1st 500+ TVS Analytics Challenge Top 200 68,000 Code Cubicle Challenge Top 250 10,000+ Adobe/Juspay National Competition Top 4000 50,000+

Academic Excellence Current Academic Standing CGPA: 8.4/10 Recent SGPA: 9.07/10 Department Rank: Top Performers Research Recognition Avishkar Research Competition Finalist Patent Filed: Generative Architecture Design Technical Sessions Conducted: 300+ Students

Certifications & Training

---

GITHUB ANALYTICS

Contribution Metrics

---

OPEN SOURCE CONTRIBUTIONS

GirlScript Summer of Code Oct 2024 - Jan 2025 | Jul 2025 - Nov 2025 Multiple merged pull requests across diverse repositories Implemented features, bug fixes, and documentation improvements Collaborated through structured PR reviews and issue tracking Maintained code quality aligned with community standards

Community Impact Technical mentorship for 300+ students Conducted sessions on AI/ML and career guidance Led engineering community initiatives at Tactiqe.in Active contributor to open-source AI/ML projects

---

EDUCATION

Bachelor of Engineering - Computer Engineering Pune Vidyarthi Griha's College of Engineering, Nashik 2022 - 2026 CGPA: 8.4/10 | Recent SGPA: 9.07/10 Consistently ranked among top performers in the department Active participation in technical clubs and research initiatives Strong foundation in algorithms, data structures, and AI systems

Core Coursework Machine Learning Deep Learning Computer Vision Data Structures & Algorithms Database Management Software Engineering

---

CONTACT & COLLABORATION

Professional Engagement

Open to discussing AI research, production ML systems, and strategic collaborations
Building cutting-edge AI applications and scalable engineering solutions

[email protected]

linkedin.com/in/atharvarahate

View Projects & Research

+91 8149812710

Mission Statement

Building AI systems that are fast, stable, and genuinely useful
Focused on production-ready solutions that scale reliably
Committed to efficient engineering and applied research