Piyush K.

Key Skills

Software

Other

Top Subject Matter

AI voice agents and LLM orchestration for call automation

Search and backend performance for AI-retrieval experiences

LLM RAG pipelines for incident RCA and structured generation

Top Data Types

Computer Code Programming

Audio

Text

Top Task Types

Computer Programming/Coding

Function Calling

Prompt + Response Writing (SFT)

Question Answering

RLHF

Text Generation

Fine-tuning

Freelancer Overview

Project: OpsEye (AI Incident Copilot) — Production-grade RAG system (assumed 2024). Brings 5+ years of professional experience across complex professional workflows, research, and quality-focused execution. Core strengths include Other. Education includes Bachelor of Technology, NIT Jalandhar (2022). AI-training focus includes data types such as Audio, Computer Code, and Programming and labeling workflows including Function Calling, Text Generation, and Fine-tuning.

Labeling Experience

Project: OpsEye (AI Incident Copilot) — Production-grade RAG system (assumed 2024)

OtherTextFine-tuning

Designed and built an event-driven RAG system (OpsEye) to automate incident analysis and root-cause detection using LLMs and contextual log retrieval. Implemented retrieval accuracy improvements via hybrid chunking and low-latency vector search. Produced structured AI outputs with citations and confidence scoring for production debugging workflows. • Designed end-to-end incident RCA pipeline using LLMs and contextual log retrieval. • Built Kafka-based ingestion pipeline to process large-scale logs in real time. • Implemented hybrid chunking (semantic + time-window + trace-aware) and Redis-based vector search. • Built LangGraph agentic workflow (embed → retrieve → rerank → generate) for automated RCA with structured outputs (citations, confidence).

2024 - Present

Founding Engineer (Backend) — AI voice automation platform (Aug 2024–Present)

OtherAudioFunction Calling

Built and deployed an AI voice automation workflow that performs real-time processing and autonomous action execution for call-based tasks. Focused on integrating voice streams and orchestrating LLM-driven steps to support post-call and in-call processing outcomes. Implemented tool invocation behavior to enable context-aware outputs and downstream actions during automation runs. • Integrated real-time voice streams from Twilio and Exotel via WebSockets for continuous AI processing. • Built agentic AI systems using MCP (Model Context Protocol) to autonomously execute actions (email, WhatsApp, scheduling). • Designed dynamic tool invocation (JRPC) for real-time and post-call AI workflows. • Developed event-driven microservices (Node.js, Kafka, Docker) to scale high-volume voice interactions.

2024 - Present

Backend Developer — AI-adjacent backend systems (Jun 2022–Aug 2024)

OtherText Generation

Developed production backend features that support large-scale AI-driven search and retrieval behaviors used by downstream AI applications. Implemented indexing and caching layers that improve retrieval and response latency for user-facing systems. While not explicitly stated as labeling work, the contributions were focused on AI-retrieval pipeline performance and structured outputs. • Built Elasticsearch-based predictor reducing search latency by 35%. • Implemented Redis caching and batch pipelines delivering notifications to 200K+ users. • Refactored and optimized backend systems to improve performance by 30%. • Engineered backend services supporting high availability and peak traffic operations.

2022 - 2024

Education

Work History