Available
Harsh Tiwari
Bengaluru, India
SeniorSoftwareEngineer
I build production Voice AI systems, LLM microservices, real-time media backends, and distributed platforms that stay fast under live customer traffic.
harsh@prod ~ voice-agent1/3
$
<1s
voice-agent response target
2M+
monthly transactions
50+
workflow node types
Experience
Work trajectory
01/03
CurrentPID_1024VOICE_AI_PROD
Senior Software Engineer
WellnessLivingBengaluru, IndiaAPR 2025 - PRESENT
Production Voice AI, LLM microservices, real-time media systems, and tenant-specific agent workflows for live customer conversations.
<1s
response-path target
5
latency stages traced
RAG
tool-backed agents
Engineered a multi-tenant Voice AI backend with RAG, structured function calling, tool schemas, and provider integrations
Built a Chinese voice-to-voice agent using LiveKit, self-deployed Qwen3-ASR/Qwen3-TTS, CUDA inference, and Gemini orchestration
Re-architected Cerebrum, the LLM microservice behind telephony agents, for sub-second streaming responses
Added service guardrails for tool execution, structured tracing, regression checks, retry-safe workflows, and read-only SQL agents
LIVEKITQWEN3GEMINIRAGKAFKAREDSHIFT
Tech Stack
Tap categories to isolate orbitsHover a category, then click a planet in orbit
Languages4 orbital bodies
PythonTypeScriptGoSQL
Backend & APIs10 orbital bodies
Node.jsNestJSREST APIsGraphQLWebSocketsOAuth 2.0Rate LimitingMicroservicesEvent-Driven ArchitectureDistributed Systems
Data & Infrastructure11 orbital bodies
PostgreSQLMySQLMongoDBRedisKafkaRedshiftAirbyteAWSDockerKubernetesCI/CD
AI/ML12 orbital bodies
LLM APIsGeminiAgentic WorkflowsFunction CallingTool ExecutionRAGVector DatabasesQwen3-ASR/TTSLiveKitCUDA/GPU InferencePrompt EngineeringAI Evaluation
Open to opportunities
Let's Build
Something.
Senior Software Engineer · Bengaluru, India
harsh@prod ~ whoami1/3
$
Location
Bengaluru
Timezone
IST +5:30
Response
< 24h
h.tiwari.dev@gmail.com · +91-7355517759