Piyush Gambhir

Software Engineer

New Delhi, Delhi, India2 yrs 4 mos experience

Highly StableAI Enabled

Key Highlights

Reduced voice agent costs by 84% through optimization.
Architected data pipelines processing over 100K records daily.
Led frontend migration to micro-frontend architecture.

Stackforce AI infers this person is a SaaS-focused Software Engineer specializing in AI and automation solutions.

Contact

Skills

Core Skills

Ai AgentsSoftware DevelopmentSoftware InfrastructureSystems Design

Other Skills

TypeScriptPrompt EngineeringTwilioTelephonyComplianceBrowser AutomationData IngestionNormalizationWeb DevelopmentCMSReactMicro-Frontend ArchitectureCRM DevelopmentWorkflow OrchestrationJavaScript

About

I build and ship production AI systems. I work on conversational AI agents for auto dealerships, focused on evaluation, response quality, and reducing cost and latency in real production environments. I have also built browser automation for legacy systems without APIs, CRM components, and large-scale data ingestion pipelines. My core interest is turning messy real-world workflows into reliable software systems that actually get used.

Experience

2 yrs 4 mos

Total Experience

2 yrs 4 mos

Average Tenure

2 yrs 4 mos

Current Experience

Spyne

2 roles

Software Engineer

Jan 2024 – Present · 2 yrs 4 mos · India · On-site

Built and operate production conversational AI agents for 50+ auto dealerships, owning evaluation pipelines, response quality tuning.
Reduced per-minute voice agent call costs from $0.87 to $0.14 (~84% reduction) through model selection, prompt optimization, and infrastructure tuning. Improved reliability, reduced hallucinations, and optimized perceived latency via streaming and turn-taking improvements.
Implemented a complete telephony layer using Twilio with full US compliance: caller ID, CNAM lookup, STIR/SHAKEN attestation, and banned-number filtering for voice; A2P 10DLC registration and campaign management for SMS.
Architected end-to-end data ingestion pipelines with browser automation (Playwright) for legacy dealer systems without APIs, processing 100K+ inventory records daily with normalization for downstream ML models.
Designed and shipped a dealer website builder with integrated CMS, enabling dealerships to self-serve site creation and content management, adopted by 30+ dealers in first quarter.
Led frontend migration from monolithic React app to micro-frontend architecture, reducing deploy times by 60% and enabling independent team shipping.
Built core CRM modules including lead management, workflow orchestration, and automated follow-up sequences for dealership sales teams.

TypeScriptPrompt EngineeringAI AgentsSoftware Development