Rushikesh – AI agent orchestration, Python, LLM

Architecture and delivery of scalable agentic AI, conversational AI, and document-intelligence solutions. Rushikesh built real-time multilingual sales-coaching experiences using LLMs, retrieval-augmented generation, voice AI, and multimodal AI avatars, integrated with enterprise workflows and cloud-native deployment infrastructure.

• Defined technical architecture and engineering strategy for client AI programs.
• Architected a multilingual conversational sales AI platform with coaching, role-play, objection handling, tool calling, and multi-turn orchestration.
• Designed a RAG pipeline using vector search, enterprise-content integration, prompt engineering, and API chaining for grounded real-time responses.
• Built voice AI infrastructure with Faster-Whisper, WhisperX, WebRTC-VAD, pyannote.audio, and low-latency inference services.
• Developed multimodal workflows and a real-time lip-synced AI avatar using OpenCV, FFmpeg, MediaPipe, and SadTalker.
• Delivered document-intelligence systems for financial documents and PDFs, including OCR, classification, structured extraction, database integration, monitoring, and deployment.

An EU-funded research initiative developing AI-driven solutions for next-generation autonomous and intelligent systems.

• Developed a LangChain-based support-routing and knowledge-retrieval agent.
• Built the user-facing experience using React and TypeScript.
• Simulated more than 1,000 user journeys and improved the solution based on conversation feedback.
• Led research and development of enterprise-scale AI architectures with 20+ industry, academic, and government stakeholders.
• Designed multi-agent systems for large-scale distributed tasks and complex workflows.

AI-powered parking-management capabilities, including automated license-plate recognition, OCR, real-time search, analytics, and operational dashboards.

• Conducted a SWOT analysis that identified performance bottlenecks and improved system efficiency by 20%.
• Developed and deployed an OCR solution that improved text-extraction accuracy by 30%.
• Integrated automated license-plate recognition with parking-management software for vehicle entry, exit, and enforcement tracking.
• Optimized Elasticsearch and FAISS search, reducing search latency by 20%.
• Analyzed data with SQL, Excel, and Tableau and created interactive dashboards for business decision-making.

Software supply-chain and license-compliance tooling to standardize Software Bills of Materials (SBOMs), validate licenses, and improve repository governance.

• Developed Python scripts to standardize SBOM generation across repositories using SPDX specifications.
• Built a platform to track software-license types and expiry dates for compliance and procurement planning.
• Integrated SBOM generation and license validation into CI/CD pipelines using Syft and Trivy.
• Contributed to the GraphBOM project by building dependency graphs and SPDX-compliant SBOMs.
• Produced reports and dashboards for stakeholders to proactively manage license risk and software governance.