SOFTWARE ENGINEER · INDIA

Hey, I’m
Ratnesh Jain

I build the systems behind conversational AI, reliable backends, and real-time data.

Explore my work ↓GitHub ↗
NOW PLAYINGBuilding agent runtimes @ FeatherHQ
Minimal lavender pixel-art workspace with a desk, monitor, plant, and computer tower. A crescent moon is visible through the window.
Small details. Big systems.EST. 2023

Work, so far.

Different teams, different problems.
Here’s what I worked on at each.

FeatherHQCURRENTSoftware Development EngineerApr 2026 — Present

Workflows for conversational agents

I’m building the infrastructure that lets a live agent follow a business process while having a natural conversation.

Making execution explicit

Designing persistent state, retries, approvals, and rules for when an agent can revisit an earlier step.

Keeping the conversation moving

Running response generation alongside data extraction so the agent can start speaking while context is still being processed.

Python / PostgreSQL / RedisEngineering notes
FeatherHQ / Apr 2026 — Present

Designing a workflow engine for live agents

Workflow execution semantics

Owned most of the design and implementation of a greenfield workflow engine for live conversational agents. Defined persistent state, validation, retries, approvals, and step-gating semantics. Establishing explicit rules for how an agent could advance or revisit a workflow as the conversation evolved.

Concurrent inference and state updates

Separated response generation and structured extraction into parallel inference paths, keeping extraction off the voice-response critical path. Coordinated tool, CRM, and knowledge-retrieval results before applying workflow-state updates. The design made the ordering of context processing and state updates explicit, accepting additional model cost to avoid serializing response generation behind extraction.

Agent memory and runtime integration

Redesigned cross-call agent memory around a relational fact store with model-driven additions, edits, and deletions after each call. Replaced vector retrieval with explicit fact management aligned to product requirements. Contributed across the shared agent runtime, PostgreSQL/Redis persistence, Celery background workloads, and OpenTelemetry/Grafana observability.

ENGINEERING NOTES / RATNESH JAIN
KovenchSoftware Development EngineerMay 2025 — Mar 2026

Voice agents, from speech to action

I built and tuned the pipeline connecting speech recognition, language models, and speech synthesis for production voice agents.

Working on the pauses between turns

Tuning endpoint timing, prompts, caching, and connection reuse while weighing response quality against latency and cost.

Making conversations useful afterward

Building a meeting-intelligence pipeline for diarized transcripts, topics, sentiment, and structured dashboard results.

LiveKit / STT · LLM · TTSEngineering notes
Kovench / May 2025 — Mar 2026

Building and optimizing a production voice pipeline

Real-time voice orchestration

Built a LiveKit-based STT/LLM/TTS pipeline for production voice agents, integrating tool execution, knowledge retrieval, and campaign workflows. Refined the agent tick loop and prompting to improve tool-use and retrieval reliability, alongside structured post-call extraction and sentiment analysis.

Latency optimization and model evaluation

Tuned silence detection, endpoint timing, prompt size, caching, connection reuse, and LiveKit behavior across the response path. Reduced configuration-dependent turn latency from approximately 2.2 seconds to 0.8–1.2 seconds. Evaluated model combinations against latency, response quality, and cost to select configurations appropriate to the workload.

Event-driven conversation processing

Built a separate meeting-intelligence pipeline that transformed recordings into diarized transcripts, normalized transcript chunks, and extracted topics and sentiment. Structured the processing stages around events and surfaced their outputs in dashboards, connecting conversation data to downstream product features.

ENGINEERING NOTES / RATNESH JAIN
DeloitteAnalystJan 2024 — May 2025

Translating rules without changing behavior

I co-developed a toolchain to migrate a proprietary IBM rule language to Java for a healthcare eligibility platform.

Inside the compiler

Implementing the parser, abstract syntax tree, and code-generation components that translated the legacy rules.

Checking that meaning survived

Validating generated Java against legacy outputs through end-to-end comparisons, alongside authentication and integration work.

Java / Parsing / ASTEngineering notes
Deloitte / Jan 2024 — May 2025

Moving business rules without changing their meaning.

The problem

A public-sector healthcare eligibility platform needed to migrate a large proprietary IBM rule-engine DSL to Java while preserving existing behavior.

My contribution

Co-developed the migration toolchain and personally implemented parser, abstract syntax tree, and code-generation components.

The result

The toolchain automated approximately 85% of legacy-rule conversion and supported completion of the full migration. Extensive end-to-end comparisons validated the generated Java against legacy rule-engine outputs.

Integration work

Built and maintained JWT/OIDC authentication flows across services and wrote integration and schema-mapping documentation for downstream teams.

ENGINEERING NOTES / RATNESH JAIN
GreenieBackend DeveloperJun 2023 — Dec 2023

Reusable candidate identity profiles

I built Rust backend services and authentication for a platform that brought multiple identity checks into one candidate profile.

Connecting verification services

Integrating identity and address checks while handling mismatched schemas, third-party throttling, and onboarding failures.

Helping integrations hold together

Contributing to Kubernetes deployments and documenting service behavior for customer development teams.

Rust / Axum / MongoDBEngineering notes
Greenie / Jun 2023 — Dec 2023

Making identity verification work across imperfect integrations.

The system

Reusable candidate identity profiles backed by Aadhaar, PAN, address-matching, passport, and related verification services.

My contribution

Built core Rust/Axum backend and authentication functionality across MongoDB and PostgreSQL. Handled third-party schema mismatches, throttling, and onboarding failures in a small startup team.

Operating the service

Contributed to a multi-service deployment on custom Kubernetes infrastructure and documented integration behavior for customer development teams.

ENGINEERING NOTES / RATNESH JAIN
AI4MBackend Developer InternJan 2023 — Jun 2023

Understanding activity on a factory floor

I built concurrent Rust consumers that turned computer-vision events into operational metrics for warehouse and automotive manufacturing deployments.

State across an event stream

Modeling loading states, process timing, safety events, and manufacturing inspection outcomes from raw detections.

Following the data end to end

Debugging across compute, queues, device I/O, scheduling, and networking in deployed environments.

Rust / Kafka / Edge systemsEngineering notes
AI4M / Jan 2023 — Jun 2023

Turning a stream of detections into an operational picture.

The system

Kafka-based industrial computer-vision pipelines across warehouse operations and automotive manufacturing. Events described PPE compliance, loading states, truck movement, process timing, and sealant application on parts.

My contribution

Built concurrent Rust consumers that converted raw inference events into business metrics. Modeled state and timing across streams, published processed outcomes for client database ingestion, and contributed to model fine-tuning and edge deployment.

Debugging the whole path

Investigated deployed behavior across compute, queues, device I/O, scheduling, and networking to understand where the system was losing time or behaving unexpectedly.

ENGINEERING NOTES / RATNESH JAIN

Open a role to see what I worked on.

Curiosity, compiled.

BUILD

PeerSync ↗

Peer-to-peer video calls. WebRTC in the browser, Java on the signaling server.

BUILD

Cloud storage ↗

A backend for uploading, managing, and sharing files, with authenticated access.

RESEARCH · CSOC 2023

Beyond passwords ↗

Co-authored research on FIDO2 authentication for constrained remote devices.