Experienced NLP expert
# Natual language processing
- 7+ years experience in information processing such as syntax analysis, machine translation, automatic speech recognition & synthesis, optical charactor recognition and 5+ years ML/DNN experience in such as ASR, OCR, NLU, NER, Q&A, chatbot, recommendation systems.
- Familiar with various NLP libraries such as Glov, Word2Vec, spaCy, gensim, Seq2seq, statsmodels, NLTK, Dialogflow, RASA, IBM Watson, BERT, GPT.
Ex: Generate plausible content for large variety of subjects. Can serve as base for marketing blogs.
Used customer's datasets and GPT-2(355M and 774M models) and fintuned model, and deploy as restAPI on cloud.
# Image processing
- Expert level knowledge in OpenCV, Dlib, Cximage
- Custom object detection, recognition, tracking, segmentation
Ex: Real-time people & vehicle counter with 95% accuracy using EfficientDet & DeepSort deployed on AWS which is used at 300 locations…
- Facial detection, recognition
Ex: facial recognition and attendance system for small company: deployed on AWS and used for automatic door management.
- Age/Gendor/Emotion detection, classification
- ALPR using openAlpr and my own offline dnn engine
Ex: Russian, chinese, hindi, USA LPR for public parks.
- Image Classification, and Segmentation using CNN, R-CNN, mask R-CNN with AlexNet, VGG, inception, imagenet and resnet
# Speech processing
- speech/speaker recognition for various language & audio types.
Ex: speach recognition system by using kaldi, deepspeech, wav2letter with under5% WER for american phoenic system.
- text synthesis and voice cloning with dnn.
Ex: speech synthesis system by using HTS, Tacotron, Voice cloning for hindi call-center.
# Data Scientist
- Solid understanding and expertise in analyzing, preprocessing, clustering, regression, visualization and maintaining large datasets.
- Web Scrapping and Data Extraction, involving all sorts of data such as structured and unstructured data from many categories of websites such as social media sites like Facebook, Instagram, e-commerce websites like amazon, homedepot, real estate websites, etc.
# Master level dnn experience with various frameworks such as tensorflow/keras, pytorch, caffe, arrayfire, flashlight, yolo, mxnet.
# Master degree on mathmatics.
- Well versed in Applied Mathematical concepts of probability, computational statistics, computational linear algebra, convex optimization, calculus, computational complexity and in their application to algorithmic trading, backtesting, time series analysis.
# Plenty of experience in deploying engines in various forms such as restful API, rpc, standarlone app, apk, docker on linux, windows, mac, cloud(AWS, GCP, Azure).
# Master level knowldege on several languages such as c/c++, python, java, Qt, PyQt, php, js.
# Good at data structure and algorithm, since having programming contestant career.
# Flexible working time on various time-zone and 40Hr/week availibility.
Freelancer was unable to solve what should have been a simple problem (for an expert in this field), and also took an extremely long time to submit anything. Very pushy to have milestone released before I or even he has properly tested his own code. Was hesitant to change/fix things at times.