Designing AI systems that operate under real-world constraints involving latency, scalability, edge deployment, realtime processing, and production-oriented architecture.

My primary interests include:

• Edge AI & Mobile Inference
• Realtime Machine Learning Systems
• Applied Computer Vision
• RAG & Knowledge Systems
• Distributed System Design
• Intelligent Automation
• Civic-Tech Infrastructure
• Financial Intelligence Systems
• Offline-First AI Platforms

I work on building practical AI systems that combine:

• low-latency inference pipelines
• scalable backend architectures
• realtime data processing
• multimodal intelligence systems
• optimization for constrained hardware
• production-oriented system design

Rather than isolated ML experimentation, I focus on end-to-end engineering pipelines involving preprocessing, inference orchestration, optimization, integration, and deployment.

Selected among the Top 30 Teams in SANKALP by Cograd.

PratibimbAI focuses on intelligent AI-assisted analytical workflows and production-oriented applied intelligence systems.

• Applied AI engineering
• Intelligent workflow systems
• AI-assisted analytical platforms
• Product-oriented system design

Graph-driven financial intelligence and scam-detection platform focused on identifying suspicious transactional relationships and anomaly patterns.

• Circular transaction analysis
• Relationship graph intelligence
• Financial anomaly detection
• Fraud-pattern exploration
• GST ecosystem analysis

Realtime deepfake and anomaly detection system optimized for mobile devices.

Designed around strict engineering constraints requiring:

• rapid preprocessing
• realtime inference
• prediction smoothing
• UI synchronization
• continuous inference cycles

The system targets sub-second processing responsiveness on mid-range mobile hardware.

• Edge AI
• Realtime ML pipelines
• Mobile inference optimization
• Computer Vision systems
• Latency-sensitive engineering
• End-to-end inference orchestration

Large-scale civic-tech platform currently under development with a production-oriented architecture.

The project is being designed around:

• scalable backend systems
• modular services
• robust API infrastructure
• distributed workflows
• automation pipelines
• production-grade deployment architecture

Exploration of retrieval-augmented generation architectures and intelligent knowledge retrieval systems.

• Retrieval pipelines
• Vector search systems
• Context-aware generation
• AI-assisted information retrieval
• Knowledge orchestration systems

AI Systems Engineering
Edge AI
Realtime Inference
Distributed Systems
Computer Vision
RAG Architectures
System Design
Multimodal AI
Backend Engineering
Mobile AI Optimization
Scalable Infrastructure
Civic-Tech
Financial Intelligence

• Winner at Techfest IIT Bombay with the project SuvyAIkth
• Selected among Top 30 Teams in SANKALP by Cograd
• 3rd Place at Vivriti Capital Competition at IIT Hyderabad
• Building AI systems focused on practical deployment constraints and production-grade engineering

Currently focused on developing:

• production-grade intelligent systems
• scalable AI infrastructure
• offline-first AI platforms
• realtime inference architectures
• constraint-aware AI deployment pipelines
• intelligent civic and financial systems

• GitHub: https://github.com/saketh1125