N

Niharika Shrivastava

About

Detail

India
Contact Niharika regarding: 
work
Full-time jobs
Starting at USD50k/year

Timeline

work
Job
school
Education
folder
Project
auto_stories
Publication

Résumé

Jobs verified_user 0% verified
  • W
    Data Scientist II
    Workforce Optimizer - AI Enabled WFM
    Apr 2024 - Current (2 years 7 months)
    • Built a time-series forecasting framework with 85% accuracy to predict hourly labour demand in 50+ hospital wards.
    • Developed an optimization framework for demand-based scheduling under uncertainty and complex constraints, managing 1K+ staff with 24/7 rescheduling in 15-minute intervals. This is built using Kotlin.
    • Prototyped 3 RAG architectures (Basic, Sentence-Window, Auto-Merging) for a company-wide chatbot. Evaluated the application using LLM-as-a-Judge on a synthetic dataset for 3 RAG-specific metrics.
    • Led project lifecycle (milestone planning, user requirements, data collection), model building (forecasting, feature engineering, optimization), and production monitoring.
  • N
    Graduate Research Assistant
    NUS Sound and Music Computing Lab,
    Nov 2023 - Jan 2024 (3 months)
    • Set up pipelines for inference, fine-tuning, and comparing open-source LLM outputs on hierarchical musical data.
  • Synapxe
    Data Science Intern
    Synapxe
    May 2023 - Aug 2023 (4 months)
    • Modelled a framework with large language models (LLMs) to generate synthetic clinical reports via prompting. • Evaluated the syntactic and semantic utility of zero-shot, few-shot, and chaining prompt techniques using LangChain to generate medical reports for 5 major domains given a public dataset (MIMIC-III). • Performed fine-tuning of WizardVicuna 13B using QLoRA and increased semantic utility by 7%.
    Experiment with different techniques to get the best performing framework.
  • Gojek
    Software Engineer II
    Gojek
    Apr 2022 - Jul 2022 (4 months)
    • Led the architectural and design decisions to build low-latency, high-throughput, scalable, and reliable backend systems on GCP. • Built real-time streaming data pipelines using Kafka, integrated data science models, and created visualisation dashboards with Prometheus for thorough performance monitoring.
    Use A/B testing to choose features loved by the customers.
  • Gojek
    Software Engineer I
    Gojek
    Jul 2020 - Apr 2022 (1 year 10 months)
    • Maintained the Order Management System and related micro-services of GoFood mobile app to ensure reliable food ordering at a scale of 1M+ orders/day.
    Monitor performance daily to meet team objectives using comprehensive dashboards.
  • S
    Undergraduate Research Assistant
    SUTDRobotics Lab,
    Jan 2020 - Jun 2020 (6 months)
    • Developed a socio-optimal vehicle path-planning framework leveraging Constraint Satisfaction Problems to provide congestion-free routes in a city; with an 84% increase in network utilization. • Proposed a novel algorithm for optimal path selection in a multi-class set-up (combination of cars, public transport, walking) to dissipate traffic in a capacity-bound transportation network; 64% decreased travel time. • Benchmarked on a diverse set of road networks from 3 metropolitan cities.
  • T
    Outreachy Intern
    The Fedora Project,
    May 2019 - Aug 2019 (4 months)
Education verified_user 0% verified
  • National University of Singapore
    Master of Computing
    National University of Singapore
    Aug 2022 - Jan 2024 (1 year 6 months)
  • Indian Institute of Information Technology
    Bachelor of Technology
    Indian Institute of Information Technology
    Jul 2016 - Jun 2020 (4 years)
Projects (professional or personal) verified_user 0% verified
  • B
    Build and Evaluate LLM RAG using Mistral
    May 2024 - Jun 2024 (2 months)
    • Built different kinds of RAG architectures (Basic, Sentence-Window, Auto-Merging) using LlamaIndex. • Created a synthetic evaluation set using Mixtral-8x7b and evaluated the RAG using a critique agent (Palm2).
  • I
    Information Retrieval and Search for Big Data
    Nov 2023 - Dec 2023 (2 months)
    • Built a custom web-scraper to collect all Cisco products' documents and their categories. Created an interactive visualization to analyse 5K+ category-product dependencies. • Constructed a FAISS index of 700K+ documents to implement search functionality for retrieving the top-k documents. • Implemented retrieval augmentation (RAG) using the index with Llama2 LLM for retrieving the top-k documents.
  • E
    Enforcing Privacy and Robustness in Machine Learning
    Aug 2023 - Nov 2023 (4 months)
    • Constructed adversarial images to make self-driving cars deviate from their original prediction using Jacobian Saliency Map Attack. Explored defences employing gradient obfuscation; created adaptive attacks to circumvent the defences. • Performed membership inference attacks for a model in the MLaaS setting. Implemented the differential privacy SGD algorithm to get meaningful privacy guarantees for the dataset.
  • S
    Singing Voice Synthesis with Avatar Generation
    Aug 2023 - Nov 2023 (4 months)
    • Generated singing audio by optimizing DiffSinger with BigVGAN vocoder and PNDM for faster inference. • Generated singing faces by fine-tuning Wav2Lip and used DeepFace augmentation to capture facial features.
Publications verified_user 0% verified
  • E
    Exploring Pillars of Trustworthy AI: Robustness and Fairness,
    Jan 2024
  • B
    Build and Evaluate LLM RAG,
    Jan 2024
  • S
    Synthetic Clinical Reports Generation using LLMs,
    Jan 2023
  • S
    Software Engineering at Practice,
    Jan 2022
  • "
    "Congestion-Aware Routing for Multi-Class Mobility-on-Demand Service,"
    Jan 2022
  • O
    Opportunities and Diversity in Tech,
    Jan 2021
  • "
    "Intrusion Detection and Attack Classification using an Ensemble Approach."
    Oct 2020
  • P
    Predicting the Traffic Jam: Congestion-Aware Routing,
    Jan 2020