Daryl Micah
Software Developer | 2.5+ years in fullstack + AI systems
Bengaluru
2+ years of work experience
Apr 20265 mosPresent
Bengaluru, Karnataka, India
· Built an AI listing-generation system using 6+ coordinated LLM agents (pydantic-ai) that can search the web and self-validate their own output — cut listing creation time from around 2 hours to under 5 minutes.
· Built a generative-image pipeline (GPT-image, background removal, scene compositing) running on async Celery + FastAPI, cutting image generation costs roughly 4x through provider fallbacks and smart caching across 3 marketplaces.
· Shipped a serverless data ingestion platform on AWS Lambda/SQS/S3, pulling in Amazon SP-API data and normalizing 8+ data types into Postgres with safe, replayable backfills.
· Delivered a full FastAPI + Next.js dashboard with secure auth and CSV/XLSX upload support, shipping ~ 400 commits across 4 services.
Jan 20251 yr 4 mosApr 2026
Gurugram, Haryana, India
• Architected a RAG pipeline (SQLCoder-8B + vector similarity) serving real-time natural-language queries via a FastAPI inference layer — improving query accuracy by 32% and replacing a manual analyst workflow used daily by 5+ team members.
• Reduced end-to-end API latency by 70% across multi-tenant Node.js/PostgreSQL services through composite indexing, query plan redesign, and elimination of N+1 patterns — directly improving SLA compliance for enterprise clients.
• Built an LLM-driven validation engine using prompt chaining and vector similarity scoring, processing
thousands of records/day autonomously and cutting manual data validation effort by 40%+.
• Designed JWT-based auth with RBAC across a multi-tenant backend, enforcing tenant-level data isolation at the middleware layer and eliminating unauthorized access incidents.
• Scaled Node.js backend from single-tenant to high-concurrency multi-tenant architecture with zero downtime migration, handling concurrent enterprise workloads reliably.
Oct 20243 mosDec 2024
Gurugram, Haryana, India
• Built FastAPI backend services for AI inference pipelines handling real-time LLM and vision model requests — optimized async request handling to cut service latency by 15%.
• Improved LLM and vision model inference throughput by 25% through quantization and structured pruning, enabling the same workload on lower-cost infrastructure.
• Trained deep learning models using Hugging Face while managing complex data pipelines.
• Optimized LLM’s & vision models with PyTorch/TensorFlow significantly improving performance.
• Deploying ML models on AWS Sagemaker, EC2 & building ML workflows using AWS Lambda.
• Developing UIs using modern frameworks like React, Next.js, TailwindCSS and FastAPI.
Jul 20237 mosJan 2024
Pune District, Maharashtra, India
• Improved data retrieval speed by 50% on an enterprise security platform by redesigning REST API response structures and eliminating redundant server round-trips.
• Delivered backend-integrated features across 3 Agile sprints with zero critical production bugs, following structured Git workflows and peer review cycles.
• Performed tasks utilizing SCRUM in an Agile Methodology environment.
Jun 20223 mosAug 2022
India
- Built datasets for Python processing, enabling efficient data analysis and manipulation.
- Developed expertise in working with a Resume Parser API, enhancing the efficiency of resume processing.
- Implemented improvements to the name parsing function of the API, significantly enhancing the accuracy and reliability of name extraction from resumes.
Daryl’s AI Stack
Education
Bachelor of Technology - BTech, Computer Engineering
The NorthCap University
20204 yrs2024
Join Daryl and 10k+ other professionals on
Tal is the app where bosses meet engineers & builders directly, without middlemen
Download Tal Boss