Khalid Sadiq Email & Phone Number

Islamabad, Pk

Poland

- Teaching Data Science Courses.- Segmentation (background removal) and Classification(tags) of clothes for more than 300 classes for ecommerce brands.

Islāmābād, Pakistan

- Chatbot (patients and doctor queries)Developed a chatbot using Rasa framework for handling patients and doctors queries for the national health system (NHS) of London for their medical app Dr.IQ- Single Sing On ( SSO )Also worked in a python team who developed SSO project for At-Tech medical products. SSO is basically to sign in once and then easily switch from one product to others without the need to enter credentials for each product. It is exactly the same concept as Google uses for its multiple products like Gmail, Drive, Docs, sheets etc.

Islāmābād, Pakistan

Lead “News Revealer” project which has a web portal and mobile app for authentication of news articles and generating news in its own wording by keeping the context and meaning of input articles same which is called paraphrasing and managed team of 5 people. 1. Crawler / Scraper for downloading Common Crawl data (news data)2. ETL on Common Crawl news data [ De-duplication(MongoDB), Filtering (Bert and Cosine), Community Detection (sister articles)]3. Trained a large NLP transformers based model on a huge amount of Common Crawl Data for re- writing a news article. Built data pipeline with generators (yield) and training pipeline from scratch in tensor-flow. 4. Deployed models on Triton Server after optimization5. RnD on Hugging-Face models and Nvidia Megatron-LM6. Collaboration with Pulses.AI and Blend360

Islamabad

- Successfully delivered McDonalds CV based use cases live and past actionable insights for Curbside, Drive-Thru, Dine-In/Take-Out, KIOSK and POS. - Worked extensively on video analytics. R&D of cutting edge algorithms of Computer Vision like face detection, face recognition, face/full-body demographics (gender&age), face emotions, people counting in crowds, object detecting and tracking, pose estimation etc.- Deployment (ML-Ops) of scale-able deep learning modules on clouds(AWS Sage-Maker) for real time business insights by processing CCTV videos.- Worked on Post-COVID19 use-cases like real time alerts of face mask detection, social distancing in public places, glove detection, detection of Corona patients using thermal cameras etc.- Managed data labeling team of 3 people. RnD for labeling tool. [Label-me, Data-Turks, Semi-supervised tools]

- TalkCheck-in Application ( lead role)This application is being used by all internal medicine patients in the US to automatically generate HPI(History of presenting illness) based on symptoms for patients & doctors. Extended it to other specialties like Psychiatry, family medicine, etc. - Domain training and Teaching Provided domain training for U.S health system and taught ML/DL major topics to newly hired members in the data science team. Motivating and Inspiring members so that they can reach their maximum potential level. Also creating a collaborative work environment among the team by helping them out in difficult stages of their work and making them trustworthy.

- Document Classification Automatic segregation of multiple types of scanned documents (Superbill, EOB, Demographics, Insurance Cards, Referrals, Face Sheet, etc. ). Transfer learning & FineTuning of all Imagenet pre- trained models (Vgg16, Inception-v3, ResNet-50, SeNet, NasNet). Finally developed a customized CNN model (TensorFlow, Keras, Pytorch)- Localization of ICD&CPT CodesTarget was to localize ICD&CPT Codes encircled or ticked by the providers on SuperBills. Synthetic data generation to cover all types of variation and used pre-trained object detection models( Yolo-v2, RCNN, Fast-RCNN, etc. )- OCR of multiple types of text imagesPerformed OCR (Pytesser-act & Google Vision API) on Superbills & EOB documents and also developed OCR APIs for Insurance and CNIC card data extraction. - Version Controlling for Data Science team on TFS using PyCharm.- Documents Similarity Unsupervised learning method to identify the similar types of documents using documents embeddings from CNN models.- Made a customized labeling tool in python for efficient labeling of different types of documents.

Fast-Nuces, Islamabad

- Worked on a Face Recognition application and developed a docker service.

Bachelor Of Science - Bs, Computer Vision | Nlp | Deep Learning | Machine Learning | Big Data | Distributed Systems

- IRIC [CNN, Tensor flow, Python] (FYP) A web application that automatically converts infrared images into RGB, color images using deep.

Fsc, Pre-Engineering, A+

Matriculation, Computer Science, A+