Open to opportunities

Shruti Singh

@shrutisingh8

PhD scholar specializing in Automatic Speech Recognition, building cross-lingual, low-resource speech and NLP solutions.

India

Message

What I'm looking for

I’m looking for a role where I can build end-to-end ML/Deep Learning systems for speech and cross-lingual low-resource settings, combining NLP (NER, OCR) with robust data pipelines to deliver reliable, real-world AI solutions.

I’m a PhD scholar specializing in Automatic Speech Recognition, with a focus on cross-lingual and low-resource speech domains. My work is driven by the goal of making speech technology effective for under-resourced, real-world settings—especially for code-mixed children’s speech.

In my current role, I conduct an extensive literature review on monolingual, bilingual, multilingual, and cross-lingual ASR systems. I also set up pipelines for collecting adult and children audio, preparing protocols for audio collection, recording at the designated location, and performing post-processing afterward.

Previously, as an Associate Data Scientist, I built NLP-based systems for information extraction from highly unstructured data. I worked with OCR outputs, used an LSTM based RNN to correct OCR-generated texts, created a hindi named dictionary via web scraping, and developed a custom Named Entity Recognition bert model for extracting person names from documents (including BFSI sector documents).

I’ve also developed models across domains, including moving object detection with moving camera setups and real-time object detection using Tiny-yolov2. Earlier projects include language translation work (Anusaaraka) with WSD rules using CLIPS and Decision Tree automation, along with VR/360-degree spatial learning research that strengthened my ability to design and evaluate data-driven studies.

Experience

Work history, roles, and key accomplishments

Current

Cross-lingual ASR System

Current

University of Petroleum and Energy Studies (UPES)

Oct 2022 - Present (3 years 9 months)

Conducting work toward a cross-lingual automatic speech recognition (ASR) system for low-resource, code-mixed children’s speech. Includes literature review, audio data collection protocol setup, recording, and post-processing.

Speech Recognition Code Mixed Speech Audio Data Collection literature review

Current

Tiny-YOLOv2 License Plate Detection

Current

University of Petroleum and Energy Studies (UPES)

Dec 2019 - Present (6 years 7 months)

Implemented Tiny-YOLOv2 for real-time object detection to detect number plates. Used an Adam optimizer to balance learning rate and momentum.

Tiny YOLOv2 Object Detection Adam Optimizer Learning Rate Tuning

Information Extraction Model

Eigenlytics

Mar 2021 - Sep 2022 (1 year 6 months)

Designed and built an information extraction approach for highly unstructured documents using NLP techniques. Implemented OCR post-processing and trained neural models including a BERT-based NER model and LSTM-based RNN text correction.

NLP OCR Post Processing BERT RNN Web Scraping Hindi Named Dictionary OCR Text Correction

Moving Object Detection (UAV)

University of Petroleum and Energy Studies (UPES)

Dec 2019 - Feb 2021 (1 year 2 months)

Conducted a literature review on moving object detection using a moving camera, specifically for unmanned aerial vehicles (UAVs).

Object Detection Camera Movement UAV literature review

Pandemic Preparedness Clustering

University of Petroleum and Energy Studies (UPES)

Mar 2020 - Jul 2020 (4 months)

Identified key factors that can improve preparedness for pandemic containment when there is no clinical cure (vaccine). Proposed a swarm intelligence-based hierarchical clustering approach to strengthen influenza containment.

Pandemic Preparedness Analysis Clustering Swarm

Head-Mounted Device VR Study

IIIT Hyderabad

Apr 2018 - Jun 2019 (1 year 2 months)

Compared desktop versus head-mounted display (HMD) virtual reality approaches for spatial learning. Also evaluated the effect of VE versus VR modes on 360-degree spatial knowledge development in an unknown environment.

Virtual Reality Oculus Rift Desktop VR

360-degree Spatial Memory

IIIT Hyderabad

Sep 2017 - Feb 2018 (5 months)

Evaluated the efficiency of a 360-degree view represented using two split 180-degree screens for developing sketched mental maps in an unknown environment based on exploration.

Mental Maps

Anusaaraka Language Translator

IIIT Hyderabad

Jul 2016 - Apr 2017 (9 months)

Worked on the Anusaaraka language translation system by creating word sense disambiguation (WSD) rules. Built both CLIPS-based rules and an automated decision-tree approach, and created concept and transliterated dictionaries to improve translation performance.

Machine Translation Word Sense Disambiguation Decision Trees Concept Dictionary Development Transliteration Dictionary CSV

Education

Degrees, certifications, and relevant coursework

University of Petroleum and Energy Studies (UPES)

Doctor of Philosophy (PhD), Automatic Speech Recognition (ASR)

2022 -

Activities and societies: Research on cross-lingual/low-resource ASR; collecting adult/children audio with protocols; building and evaluating ASR models.

PhD research scholar specializing in Automatic Speech Recognition, focusing on cross-lingual and low-resource speech domains, including code-mixed children’s speech. Conducts literature review, audio data collection setup, and model building.