Industrial manual RAG
Retrieval-augmented chat system for querying industrial equipment manuals, connecting language-model responses directly to source documentation with zero external egress.
I'm Eklavya Ghodake — I build software for the factory floor: real-time telemetry, distributed streaming pipelines, and AI that reads engineering drawings.
Software R&D Lead / Manager at Yantramedhavi India, architecting event-driven telemetry microservices, zero-loss disaster recovery, and edge services for industrial manufacturing.
Creator of open-source developer tools including MQTT-GO (concurrent Go client) and CTQ-Extraction (AI drawing extraction skill).
less database storage after moving to event-driven + aggregated time-series
data loss recovering corrupted production MongoDB clusters
RAM footprint of MQTT-GO (vs Electron-based tools)
platform uptime across enterprise deployments
High-throughput industrial telemetry, lightweight network debugging platforms, and vision-based drawing extraction.
Industrial telemetry platform cutting time-series DB storage by 70%, architecting zero-loss disaster recovery, and streaming live factory dashboards.
Ultralight concurrent MQTT client with time-travel debugging, multi-broker streams, and sub-100MB active RAM footprint.
AI agent skill extracting Critical-To-Quality parameters, GD&T feature control frames, and verbatim tolerances from engineering drawings into validated JSON.
Bridging factory-floor physical domain expertise with high-throughput streaming pipelines, low-latency microservices, and applied AI.
I design and scale high-throughput backend microservices, real-time event-driven architectures, and responsive operational dashboards. My engineering background in mechanical systems gives me native fluency in machine kinematics, PLC telemetry, and sensor protocols that generic software engineers often struggle to decode.
Impact is backed by production metrics: re-architecting continuous time-series models to cut storage costs by 70%, engineering automated archive-based disaster recovery pipelines with zero data loss, and sustaining 99.9%+ uptime SLAs across mission-critical enterprise installations.
Leading industrial software architecture, event-driven streaming pipelines, and mission-critical production systems.
Delivered technical product architecture and live platform demonstrations of real-time data streaming systems to C-suite executives and plant heads, articulating scalability, system design, and operational ROI.
Presented real-time data streaming architectures, low-latency edge computing, and secure systems integration to defence-sector stakeholders and technical leaders.
IACSD, CDAC Pune
RMD Sinhgad College of Engineering
Walchand College of Engineering
Grounded in practical engineering: making industrial manuals searchable, running models locally, and turning unstructured drawings into validated structure.
Retrieval-augmented chat system for querying industrial equipment manuals, connecting language-model responses directly to source documentation with zero external egress.
Optimized inference workflows using Ollama and llama.cpp on edge hardware and local clusters, guaranteeing tight enterprise data boundary compliance.
Multimodal agent pipeline extracting Critical-To-Quality tolerances and 14 ASME Y14.5 / ISO 1101 GD&T symbols from engineering drawings into JSON schemas.
Experimental agent skills, local RAG architectures, MCP servers, and lightweight IoT protocol utilities.
AI agent skill for refactoring legacy UI and building enterprise interfaces with shadcn/ui, Obsidian Dark & High-Contrast Light, capsule geometry
MCP server connecting Antigravity, Claude and Cursor to DeepSeek Harness for $0-cost autonomous coding and dual-agent verification
On-premises RAG engine with local LLM inference, zero cloud data egress, and source-grounded citation
Core competencies across languages, real-time telemetry protocols, and automated drawing extraction.
Whether you want to discuss distributed architectures, industrial IoT telemetry, or applied AI engineering—my terminal and inbox are open.
Leading high-throughput distributed microservices, factory-floor streaming pipelines, and event-driven architectures.