Vaibhav Jain

Dual-model AI microservice platform that dynamically tailors resumes to job descriptions by coupling a fine-tuned T5 model for creative text rewriting with Google's Gemini API for keyword extraction and relevance scoring.

• Built a FastAPI service exposing REST endpoints that transform raw resume input into structured, recruiter-ready content with high relevance and low latency.
• Orchestrated dual AI models: fine-tuned T5v1_v1 for bullet point rewriting and Gemini API for keyword extraction and ATS match scoring.
• Engineered end-to-end evaluation pipelines to validate output fidelity and precision before delivery to the React/Vite interface.

Multi-agent conversational financial guidance system built for the Economic Times Gen AI Hackathon, converting user dialogue into a dynamic financial profile and surfacing personalized market insights.

• Designed multi-agent workflows that parse conversational inputs to infer user financial goals, risk appetite, and portfolio stages in real time.
• Synthesized complex market signals and ET ecosystem content into plain-language actionable recommendations and smart alerts.
• Architected modular service boundaries separating agent reasoning, user context state, and partner service routing.

Autonomous edge computing pipeline for resource-constrained devices (Broadcom BCM2837 SoC) that extracts real-time information-theoretic features and dynamically selects Pareto-optimal compression algorithms.

• Ingests live sensor streams (DHT22 GPIO, CSI camera module, SoC hardware telemetry) with minimal latency on the edge node.
• Computes real-time entropy and statistical features to forecast device resource availability and network bandwidth.
• Dynamically selects Pareto-optimal compression algorithms under strict data reconstruction error and energy consumption constraints.