Amal Singh
AI & ML Engineer @ Benori | Production Agentic Systems & RAG at Scale (LangGraph, Multi-Agent, Google ADK)
Gurugram
3+ years of work experience
Apr 20263 mosPresent
Mar 20251 yr 2 mosApr 2026
Gurugram, Haryana, India
Driving innovation by building and integrating end-to-end AI systems into business workflows, optimizing operations, and enabling data-driven efficiency.
Sep 20247 mosMar 2025
New Delhi, Delhi, India
Project 1: Classification of Eye Fundus Images for Multiple Eye Related Diseases (Ongoing)
- Developed a data augmentation pipeline for a Fundus Image Dataset (45,000+ images) using exposure adjustments, rotation, and flipping to improve eye disease classification.
- Implemented a SOTA model for Diabetic Retinopathy detection, achieving 96%+ recall and 87% specificity, surpassing competitors by 14% in recall.
- Trained DenseNet, InceptionV3, and ResNet50 on grayscale fundus images, achieving 92.09% accuracy and 94.47% recall in binary classification.
- Built a SOTA model for Cup-to-Disc ratio detection in Glaucoma screening, manually annotating 350+ images and achieving 0.994 mAP and 92.88% recall.
Project 2: Video-Based Body Vitals Prediction (Ongoing)
- Collected data from 39 subjects across multiple activity states, capturing BVP, facial video, and blood pressure data.
- Implemented MLFlow for model monitoring and deployment, achieving ±5 BPM heart rate estimation accuracy.
- Designed a custom weighted loss function (shape-based + Negative Pearson loss) to enhance BVP prediction.
- Optimized inference time, reducing 30s video processing from 25s to <9s (65% improvement).
Project 3 : Streamlit Dashboard for deployment in low connectivity areas
- Created a modular, user-friendly Streamlit application for site engineers to analyze soil sample data in low-internet connectivity environments.
- Packaged the application as a standalone executable (.exe) for universal system compatibility, enabling seamless data visualization and quality assessment.
Jun 20243 mosAug 2024
Hyderabad, Telangana, India
- Automated sales outbound workflows, linking Google Sheets with Apify for profile scraping, integrating Apollo.ai for lead email extraction, and generating personalized email sequences stored as drafts in the customer’s Gmail inbox. This solution saved the client over $15,000+ in sales outbound costs, significantly reducing dependency on manual outreach.
- Developed agentic workflows for HR recruitment and candidate screening, implementing an AI-generated resume relevancy score and integrating OpenAI Whisper for voice-based Q&A, enabling human-like conversations for enhanced candidate interactions.
- Designed and developed Shopify Liquid templates, integrating a B2C customer's B2B vertical with features such as account creation, promotional banners, forms, user authentication, cataloging, and automated promotional email workflows.
Apr 20243 mosJun 2024
Gurugram, Haryana, India
- Streamlined resource allocation, eliminating bottlenecks and redundancies, resulting in a savings of ~ 2 Days per sprint.
- Led the development of a multi-user chatbot, leveraging advanced LLMs to facilitate digital mediation, achieving 90% accuracy in scenario-based response relevancy tests.
- Developed a multi-user chatbot, leveraging advanced LLMs to facilitate digital mediation, automating the conflict resolution upon getting relevant context from private conversations with multiple parties.
Skills: Python, Prompt Engineering, RAG, Tailwind, Supabase, Langchain, Langgraph
Oct 20217 mosApr 2022
Hyderabad, Telangana, India
Contributed to an app and web admin panel using SQL, enhanced search and chatbot with NLP, and ReactJS.
Education
Postgraduate Diploma, Data Science (AI/ML)
Plaksha University
20231 yr2024
Bachelor of Business Administration - BBA, Business Management, Business Analytics
GITAM Deemed University
20193 yrs2022
Join Amal and 10k+ other professionals on
Tal is the app where bosses meet engineers & builders directly, without middlemen
Download Tal Boss