# Vishvdeep Dashadiya (AI with VD), Lead AI Engineer

> I design, build, and deploy intelligent products with rigorous engineering, clear metrics, and a calm delivery cadence. From agentic workflows to resilient MLOps, we move from prototype to scale without chaos.

Portfolio, services, case studies and technical blog. Core focus: Agentic AI, real-time ML systems, and cloud-native infrastructure. Areas of expertise: Machine Learning, Generative AI, LLM Engineering, Agentic Systems, MLOps, RAG, LangGraph, FastAPI.

## About

I design, build, and deploy intelligent products with rigorous engineering, clear metrics, and a calm delivery cadence. From agentic workflows to resilient MLOps, we move from prototype to scale without chaos.

- Website: https://aiwithvd.com
- Email: contact@aiwithvd.com
- LinkedIn: https://www.linkedin.com/in/vishvdeep/
- GitHub: https://github.com/aiwithvd

## Expertise

- Machine Learning: Production-grade ML systems built for measurable business impact.
- GenAI Engineering: LLM-powered products that scale from idea to deployment.
- MLOps: Operational excellence for models in the wild.

## Experience

- Lead AI Engineer @ WeyBee Solutions Pvt Ltd (Aug 2026 - Present): Leading AI engineering efforts, architecting intelligent systems and driving end-to-end delivery of production AI products.
- Senior Software Engineer - AI/ML @ BankBenchers (Nov 2025 - Jul 2026): Leading AI/ML initiatives with a focus on agentic systems, cloud-native deployment, and scalable production workflows.
- Board of Studies - External Industry Expert @ Atmiya University (May 2024 - Present): Advising academic curriculum and industry alignment for AI and data science programs, bridging classroom outcomes with real-world needs.
- Senior AI Engineer @ hypREspace (Jul 2025 - Oct 2025): Delivered AI consulting and platform capabilities, emphasizing AWS-based deployments and applied generative AI solutions.
- Senior Software Engineer @ 7Span (2024 - 2025): Spearheading GenAI and ML applications in Python, building scalable pipelines, and integrating AI solutions into enterprise systems.
- Data Science Engineer & Manager @ fxis.ai (2022 - 2024): Led ML engineering teams through delivery, focusing on data quality, model optimization, and stakeholder alignment.
- Software Engineer @ Accenture (2019 - 2022): Drove feature development, impact analysis, and cross-team delivery in Agile environments.

## Services

- [AI-Assisted Custom Software Development](https://aiwithvd.com/services/ai-assisted-custom-software-development): Web, API and data products built with AI-assisted engineering and real code review.
- [Agentic Workflow Development](https://aiwithvd.com/services/agentic-workflow-development): LangGraph agents with tools, memory and evaluation gates that hold up in production.
- [LLM Fine-tuning, Inference and Deployment](https://aiwithvd.com/services/llm-fine-tuning-inference-deployment): SFT and LoRA fine-tuning, evaluation harnesses, and serving that fits the budget.
- [MLOps](https://aiwithvd.com/services/mlops): Pipelines, monitoring, versioning and cloud deployment for models in the wild.
- [RAG Applications](https://aiwithvd.com/services/rag-applications): Retrieval systems with hybrid search, reranking and traceable answers.
- [Document Intelligence](https://aiwithvd.com/services/document-intelligence): OCR plus LLM extraction pipelines that turn documents into structured data.
- [Voice AI](https://aiwithvd.com/services/voice-ai): Real-time voice assistants with explicit local and online processing boundaries.
- [AI Strategy and Consulting](https://aiwithvd.com/services/ai-strategy-consulting): Roadmaps, feasibility checks and architecture reviews before you commit budget.

## Case Studies

- [Designing Secure Agent Transaction Infrastructure](https://aiwithvd.com/case-studies/secure-agent-transaction-infrastructure): A protocol-neutral reference design for giving AI agents bounded authority, exact approvals, isolated credentials, safe outcome handling and auditable transaction evidence. (Independent Product R&D; Evidence: Target; Results: Design outputs)
- [hypREspace: Cutting Token Costs 90% While Raising Tool-Calling Accuracy to 95%](https://aiwithvd.com/case-studies/hyprespace-agent-cost-optimization): Dynamic prompt injection, a gold evaluation set and Amazon Nova fine-tuning cut a live agent's token spend by 90% and lifted tool-calling accuracy from 75.8% to 95%. (Client Delivery: hypREspace; Evidence: Measured; Results: Measured outcomes)
- [hypREspace: Rearchitecting NL-to-SQL into an Agentic Analytics Engine](https://aiwithvd.com/case-studies/hyprespace-nl-to-sql-agent): A brittle multi-hop prompt chain became one observable LangGraph ReAct agent with a custom MCP layer over the existing FastAPI backend and Claude on Bedrock. (Client Delivery: hypREspace; Evidence: Production; Results: Production outcomes)
- [AI Research Platform for a Regulated Accounting Profession](https://aiwithvd.com/case-studies/regulated-accounting-research-platform): End-to-end MVP architecture for cited, verified tax research answers: LangGraph agent, hybrid RAG, a claim verification gate and region-locked multi-tenant infrastructure. (Client Delivery: Compliance-focused accounting technology startup (under NDA); Evidence: Target; Results: Design targets)
- [Agentic Question Generation for a Family Connection App](https://aiwithvd.com/case-studies/family-app-question-generator): A LangGraph agent with agentic RAG generates personalised daily storytelling prompts as strict JSON, served to a live mobile app through FastAPI on AWS. (Client Delivery: Premium family connection platform (under NDA); Evidence: Production; Results: Production outcomes)
- [Document Extraction Pipeline: Scaling AI Workloads with FastAPI and Celery](https://aiwithvd.com/case-studies/document-extraction-pipeline): An async-first extraction service: FastAPI accepts uploads, Celery workers run OCR and LLM extraction, and typed schemas turn invoices, legal documents and ESG reports into JSON. (Self-Built Project; Evidence: Demonstrated; Results: Demonstrated capabilities)
- [Self-Hosted Voice AI: A Local-First Hybrid Pipeline](https://aiwithvd.com/case-studies/self-hosted-voice-ai): A local-first, hybrid voice pipeline behind LiveKit WebRTC: VAD, STT, and LLM run locally, while default Edge-TTS sends response text to Microsoft Edge's online TTS service. (Self-Built Project; Evidence: Demonstrated; Results: Demonstrated capabilities)

## Projects

- [Secure Agent Transaction Infrastructure: Bounded authority for agents that can transact](https://aiwithvd.com/projects/secure-agent-transaction-infrastructure): Tool access does not by itself establish transaction authority. This protocol-neutral target reference architecture turns agent proposals into bounded, policy-checked, transaction-bound, credential-isolated and auditable external actions.
- [Adaptive RAG: Toggle-Driven Hybrid Retrieval With Skills](https://aiwithvd.com/projects/adaptive-rag): A self-hosted LangGraph and FastAPI RAG service for a regulated finance and legal practice. Vector search always runs; allowlisted web search is added only when the request toggles it on, never by an LLM router. Four self-reflection gates with hard-capped retry loops decide whether an answer ships. Slash-invocable skills run as isolated sub-agents over the same graph, and opt-in Langfuse tracing plus async DeepEval scoring observe every run without adding latency.
- [Fine-Tuned Tool-Calling Agent: A Small Model Taught to Use 40+ Tools](https://aiwithvd.com/projects/tool-agent-finetune): A LangGraph ReAct agent over a custom MCP backend exposing 40+ tools, with the decision model replaced by a fine-tuned Gemma 3 4B. Successful multi-turn tool trajectories from production were filtered and reformatted into an SFT dataset, then used to train Gemma 3 4B with QLoRA — cutting cost and latency while making tool-call formatting far more reliable.
- [Document Extraction Pipeline: Turn Documents into Structured Data with AI](https://aiwithvd.com/projects/document-extraction): A production-ready FastAPI service that extracts structured data from PDFs and images using OCR and LLM technology. Features async processing, JWT authentication, and support for multiple document types including invoices, legal documents, and ESG reports.
- [LangChain DeepAgent: Production-Ready AI Agent with Specialized Skills](https://aiwithvd.com/projects/deepagent): A FastAPI service powering a LangChain DeepAgent backed by Ollama llama3.2:3b. Features four specialized skills for complex reasoning tasks, SSE streaming for real-time responses, Redis rate limiting, and multi-turn session memory.
- [BoxCricket Umpier: AI-Powered Cricket Scoring for Gully Cricket](https://aiwithvd.com/projects/boxcricket): A Next.js 14 mobile-optimized cricket scoring application for box cricket and gully cricket matches. Features ball-by-ball scoring, intelligent rules engine, live statistics, and smart undo functionality.
- [Voice AI Demo: Local-First Hybrid Conversational Voice AI Assistant](https://aiwithvd.com/projects/voiceai): A local-first, hybrid conversational voice AI assistant powered by LiveKit Agents. VAD, STT, and LLM run locally with Apple Silicon acceleration; default Edge-TTS sends response text to Microsoft Edge's online text-to-speech service.

## Blog

- [Jev in LangGraph: Bounded Answers for Bounded Questions](https://aiwithvd.com/blog/jev-langgraph-bounded-answers): Jev is a model that never generates text — it answers typed questions with calibrated probabilities. Where that fits in a LangGraph agent, with code, and where I would not use it.
- [Where Secure Agent Transactions Fit: A Use-Case Map](https://aiwithvd.com/blog/where-secure-agent-transactions-fit-use-case-map): A decision framework for applying governed transaction controls across procurement, travel, paid APIs, recurring operations, and service purchasing.
- [What Exactly Did the Human Approve?](https://aiwithvd.com/blog/what-exactly-did-the-human-approve): A practical model for turning an approval click into typed, transaction-bound authority that deterministic controls can verify.
- [“Uncertain” Is a State, Not an Error](https://aiwithvd.com/blog/uncertain-is-a-state-not-an-error): Why ambiguous transaction outcomes must pause blind retries and move through evidence-led reconciliation.
- [One Authorization, One Commit Boundary](https://aiwithvd.com/blog/one-authorization-one-commit-boundary): How to consume approval, reserve constraints, issue scoped authority, and hand off evidence without pretending an external side effect is atomic.
- [Keeping Provider Credentials Outside the Agent](https://aiwithvd.com/blog/keeping-provider-credentials-outside-the-agent): A target design for workload identity, credential brokering, short-lived grants, audience restriction, and isolated execution.
- [From Request to Receipt: A Procurement Transaction Walkthrough](https://aiwithvd.com/blog/from-request-to-receipt-a-procurement-transaction-walkthrough): A fictional purchase traced through identity, policy, approval, scoped execution, uncertain outcomes, reconciliation, and final evidence.
- [Designing Spend Policies for Autonomous Buyers](https://aiwithvd.com/blog/designing-spend-policies-for-autonomous-buyers): How deterministic spend policy combines action limits, budgets, supplier rules, validity, velocity, risk, and separation of duties.
- [Building a Governed Procurement Agent](https://aiwithvd.com/blog/building-a-governed-procurement-agent): An illustrative procurement architecture that separates agent planning, spend policy, exact approval, isolated purchasing, and receipt evidence.
- [Tracing and Scoring a RAG Pipeline Without Slowing It Down](https://aiwithvd.com/blog/tracing-and-scoring-rag-without-slowing-it-down): Every run of the pipeline becomes a nested Langfuse trace, and a judge model scores it with DeepEval. Neither adds a millisecond to the response, because scoring starts after the client already has the answer. The wiring: a per-request callback, a fire-and-forget task that cannot fail the request, a judge decoupled from the graph LLM, and an offline golden set for the metrics live traffic cannot compute.
- [Skills as Sub-Agents: One Graph, Many Personas](https://aiwithvd.com/blog/skills-as-sub-agents-one-graph-many-personas): A skill is a folder with two files. SKILL.md becomes the system prompt, skill.yaml declares which sources the run may read, and the shared LangGraph pipeline runs unchanged on an isolated thread. Slash-invocable, like a command. Here is how a domain persona parametrizes four points of one graph without forking it, and the allowlist bug a code review found along the way.
- [It's Not Adaptive RAG: Why I Let the Human Choose the Route](https://aiwithvd.com/blog/its-not-adaptive-rag-toggle-driven-hybrid-retrieval): I built a RAG service for a regulated finance and legal practice and called it adaptive-rag. It isn't. There is no LLM router picking a path. Vector search always runs, web search is added only when a person flips a toggle, and four self-reflection gates with hard-capped loops decide whether the answer is good enough to ship. Here is why that design beat the classic one.
- [Fine-Tuning a Tool-Calling Agent: SFT + QLoRA on Gemma 3 4B](https://aiwithvd.com/blog/fine-tuning-tool-calling-agent-sft-qlora-gemma): How I turned a prompt-driven ReAct agent over a 40+ tool MCP backend into a fine-tuned one — collecting multi-turn tool trajectories, building the SFT dataset, training Gemma 3 4B with QLoRA, and serving it back behind the agent loop. The case study that ties the whole fine-tuning series together.
- [LoRA & QLoRA: Fine-Tuning a Model That Doesn't Fit on Your GPU](https://aiwithvd.com/blog/lora-qlora-fine-tuning-one-gpu): Full fine-tuning a 7B model needs ~112GB before you load a batch. LoRA trains ~1% of the weights; QLoRA squeezes the frozen base into 4 bits so a large model fits on a single card. Here's how both work — and the four pieces of QLoRA interviewers always probe.
- [SFT: What the Model Is Actually Predicting (and the Mask That Decides If It Works)](https://aiwithvd.com/blog/sft-what-the-model-predicts): Supervised fine-tuning looks like 'teach the model the answer.' Mathematically that's not what happens. It's the same next-token prediction as pretraining — and two quiet details, response masking and the chat template, decide whether your fine-tune works or silently rots.
- [Inside a Transformer Block: Why Where Knowledge Lives Decides Where You Fine-Tune](https://aiwithvd.com/blog/inside-a-transformer-block): Most people learn q_proj, k_proj, v_proj, o_proj and think that's the whole transformer. It isn't. A block has two machines — attention and the MLP — and knowing which does what is the difference between a LoRA config that learns your domain and one that doesn't.
- [The Two Axes of Fine-Tuning: A Mental Model That Stops the Confusion](https://aiwithvd.com/blog/two-axes-of-fine-tuning): LoRA, SFT, QLoRA, DPO, PPO, GRPO — they all blur together until you see they live on two independent axes. One decides how you touch the weights, the other decides what signal you train on. The map I use before every fine-tuning project.
- [Using Mem0 for User Preference Memory and Context Switching in AI Agents](https://aiwithvd.com/blog/mem0-user-preference-memory-context-switching): How to store user preferences in Mem0, retrieve them by context tag before each response, and handle context switching when users shift between different tasks. Practical patterns with LangGraph integration.
- [Building a 4-Tier Verification Layer for LLM Outputs in Regulated Domains](https://aiwithvd.com/blog/building-verification-layer-llm-outputs): How I built a verification pipeline that catches hallucinated citations before they reach production. NLI cross-encoder, Claude judge, hard-fail gates, and the economics of getting it right at $0.05 per query.
- [The Rise of the Autonomous Company: Hermes, OpenClaw, and the Paperclip Revolution](https://aiwithvd.com/blog/the-rise-of-the-autonomous-company-hermes-openclaw-and-the-paperclip-revolution): A comparative analysis of the leading AI agent frameworks in May 2026: Hermes' self-improvement, OpenClaw's massive ecosystem, and Paperclip's orchestration.
- [Building a Local-First Voice AI Assistant with LiveKit Agents](https://aiwithvd.com/blog/building-voice-ai-assistant-livekit): How I built a real-time, hybrid voice AI: VAD, STT, and LLM run locally, while default Edge-TTS sends response text to Microsoft Edge's online TTS service.
- [Prompt Engineering vs RAG vs Fine-Tuning: The Decision Framework I Actually Use](https://aiwithvd.com/blog/prompt-rag-finetune-decision-framework): Stop guessing which LLM technique to use. Data-driven decision framework with ROI analysis, cost comparisons, and real production case studies from Document Extraction Pipeline and DeepAgent.
- [Evaluating Generative AI in Production: Metrics Beyond 'Correct' and 'Incorrect'](https://aiwithvd.com/blog/evaluating-generative-ai-production): Stop eyeballing LLM outputs. Learn production-grade evaluation: LLM-as-judge, deterministic checks, RAG faithfulness, and A/B testing frameworks that actually work.
- [Memory Management for AI Agents: Context Window Optimization Without Token Bankruptcy](https://aiwithvd.com/blog/ai-agent-memory-management): Stop letting context windows bankrupt your AI budget. Learn sliding window, summarization, and vectorized memory strategies with real cost savings from production deployments.
- [From Document Processing to LLM Resilience: Patterns That Scale](https://aiwithvd.com/blog/resilient-llm-api-patterns): Building on Document Extraction Pipeline's Celery/Redis foundation, learn to extend async patterns to LLM-specific resilience: circuit breakers, multi-provider fallbacks, and token bucket rate limiting.
- [Production-Grade LLM System Architecture: From Notebook to 10k RPM](https://aiwithvd.com/blog/production-llm-system-architecture): Learn how to design decoupled LLM systems that handle 10,000 requests per minute. Covers async queues, caching strategies, RAG integration, and the three biggest bottlenecks you'll face.
- [OpenClaw: A Self-Hosted AI Assistant with Ollama, Telegram & Discord](https://aiwithvd.com/blog/openclaw-self-hosted-ai-assistant-guide): Install OpenClaw, wire Anthropic/OpenAI/Google or a local Ollama model, control it from Telegram and Discord, extend it with skills.sh, and turn it into a business gateway for lead qualification, customer support, and agent-ecosystem management.
- [Building a Production-Ready LangChain DeepAgent with SSE Streaming](https://aiwithvd.com/blog/building-langchain-deepagent-sse-streaming): How I built a production AI agent API using LangGraph, Ollama, and FastAPI with real-time SSE streaming, Redis rate limiting, and multi-turn session memory.
- [Getting Started with LLM Agents in Production](https://aiwithvd.com/blog/getting-started-with-llm-agents): A practical guide to building and deploying LLM-powered agents using LangChain and FastAPI, with lessons from real-world implementations.
- [Building a Production-Ready Document Extraction Pipeline with FastAPI and MinerU](https://aiwithvd.com/blog/building-document-extraction-pipeline-fastapi-ollama): How I built a high-performance document extraction system using FastAPI, Celery, MinerU, and LLMs. Learn the architecture, challenges, and lessons from deploying AI-powered document processing at scale.

## Other

- [Blog index](https://aiwithvd.com/blog)
- [RSS feed](https://aiwithvd.com/rss.xml)
- [Sitemap](https://aiwithvd.com/sitemap.xml)