AI Deploy Network / Builder
AI Engineer
Published evidence
6 AI Deployments
6 publicly identified
2 Case Studies
Published implementation narratives
01
I build LLM systems and the machinery that keeps them honest: agentic orchestration that runs unattended, retrieval that actually grounds an answer, and the evaluation harnesses that tell you when a model has quietly gotten worse.
Most of my work starts from the same observation. ML systems rarely fail with a crash, they fail quietly. A recommendation drifts across sessions, a detector's per-class recall collapses after a retrain, an agent reasons confidently from a false premise. So I spend as much time on benchmarks, failure analysis and supervision as I do on the models themselves.
In production I built an agentic stock intelligence platform on GPT-4o with a custom MCP orchestration layer, shipped as a containerised five-service system on GCP, producing investment signals across 24 equity buckets daily with no human in the loop. Alongside it I fine-tuned an RT-DETR detector for real-time drone perception to over 90% mAP at under 33 ms per frame, and built the benchmark suite that regressed model quality version over version.
Independently I build agent reliability and retrieval tooling: AgentFuse, an open-source circuit breaker that detects tool loops, goal drift, logic traps and runaway spend in long-running agents, and a hybrid retrieval system that lifted nDCG@5 from 0.645 to 0.728 on a held-out split by repairing extraction rather than reranking.
B.Tech in Computer Science and Engineering from IIIT Vadodara. Currently an AI Engineer at BBI Inc. I work on agent orchestration, retrieval and model evaluation, and I am open to implementation work in those areas.
02
Implementation records and publicly available work, where published.
6 AI Deployments delivered, including 6 publicly identified implementations available to inspect.
03
Implementation domains listed on this profile. Evidence-supported items are identified separately below.
Agentic Workflow Automation
Autonomous Agents
Document Intelligence
Multi-Agent Systems
RAG & Knowledge Retrieval
04
Real AI implementations demonstrated through evidence.
Local engineering copilot that turns tickets into reviewable implementation briefs, linking existing behavior to code and proposed changes to requirements and tests. Includes human clarification, deterministic verification and retrieval evaluation.
Software & SaaS · AI Copilot · Internal Developer Tooling
View DeploymentDeployed support copilot for an e-commerce and subscription demo: Claude tool-calling, policy RAG, reliability scoring, human review, and resilient background refund workflows with simulated commerce integrations.
E-Commerce · Customer Support Agent · RAG System
View DeploymentRed team and blue team wired into a loop. A knowledge graph of payment-fraud vectors feeds an attack simulator, which feeds an XGBoost detector. Missed detections return as harder variants, so the detector's own blind spots choose the next attacks.
Payments · AI Evaluation Framework · Fraud Detection
View DeploymentGive it a PDF and a natural-language query and it returns the pages and section headings that answer it. EmbeddingGemma vectors fused with BM25, evaluated on 45 graded queries with 12 held out: nDCG@5 0.645 to 0.728, hit@3 82% to 93%.
Cross-Industry / Horizontal · AI Evaluation Framework · Question Answering · RAG System · Semantic Search
View DeploymentOpen-source reliability layer above the agent execution graph. It reads telemetry frameworks already emit and trips a breaker on tool loops, goal drift, logic traps or runaway spend, then steers the agent back or escalates to a human.
Cross-Industry / Horizontal · AI Agent · AI Evaluation Framework · AI Orchestration
View DeploymentProduction agentic system that autonomously produces investment signals across 24 equity buckets daily with no human in the loop, via a custom MCP orchestration layer over four tools and a vector memory tracking portfolio drift across sessions.
Capital Markets · AI Agent · AI Orchestration · Multi-Agent Systems · RAG System · Task Automation
View Deployment05
Evidence-supported means linked to a published Deployment or Case Study; other capabilities are Builder declared.
06
Technologies listed by the Builder, distinguished by published evidence.
07
Published Deployments support the listed counts. Other industries are Builder declared.
2 published Deployments
1 published Deployment
1 published Deployment
1 published Deployment
1 published Deployment
Builder declared
Builder declared
08
Published narratives of implementation and results.
Cross-Industry / Horizontal · AI Evaluation Framework · Question Answering · RAG System · Semantic Search
An independent retrieval project, live on Cloud Run. Evaluated on 45 graded queries with 12 held out: nDCG@5 improved from 0.645 to 0.728 and hit@3 from 82 percent to 93 percent. Three standard ranking upgrades were built, measured against the held-out split, and rejected.
Read case studyInvestment Banking · AI Agent · AI Orchestration · Multi-Agent Systems · RAG System · Task Automation
DigitalxCode needed defensible daily equity coverage across two dozen buckets without scaling headcount. Built a GPT-4o agent over a custom MCP orchestration layer, shipped as five containerised services on GCP Compute Engine, running unattended every night behind a live dashboard.
Read case study09
A closer look at how this profile and its evidence are reviewed.
Community Member
AI Case Studies verified
Published evidence is reviewed against the AI Deploy Network Evidence Standard.
Verified Builder
Independently reviewed by AI Deploy Labs against the platform Evidence Standard.
The platform does not independently audit client systems or financial records. How this is verified
Digitalxcode
Agentic AI • IT Services & Systems Integration
Built an AI automation agents for daily stock analysis, curating the best list of stocks across different countries, and different time horizons.
Coursera Machine Learning
Coursera
Issue date not specified
NVIDIA Deep Learning Institute Certification
NVIDIA
Issue date not specified
Tell Navin what you are trying to automate and the outcome you need. Your message goes directly to the builder, with a copy kept in your Organisation Workspace.
