CV

Education

• B.S. in Neuroscience and Anthropology, University of Michigan, 2011
• Postgraduate Studies - Economics, Harvard Business School, 2015
• Advanced Postgraduate Studies (Non-degree seeking) - Statistics and Applied Mathematics, North Carolina State University, 2017 - 2020
• Masters of Science - Applied Mathematics, Johns Hopkins, 2021

Work and Research experience

• July 2022 - October 2022: Data Scientist
  • Acre Labs
• November 2021 - June 2022: Digital Service Expert - Data Scientist
  • United States Digital Service (Executive Office of the President)
    Contributions to the Justice40 Initiative:
    1. Implemented and presented statistical analyses - leveraging spatial statistics and non-parametric inference - to inform policy decision-making in identifying disadvantaged communities that are experiencing environmental injustice
    2. Collaborated with federal agency experts and state-level stakeholders to develop data pipelines that integrated new public health, economic, and environmental measures into the environmental justice screening tool
       Contributions while detailed at the Centers for Medicare and Medicaid (CMS):
    3. Developing and evaluating comparative predictive models for opioid-related hospitalizations based on prior contraindications and prescription consumption
    4. Developing mixed effects models to measure differences in mental healthcare consumption in the Medicaid population. Evaluated models in various state jurisdictions to revise quality measures for mental healthcare services
    5. Serving as consultative subject matter expert for projects that involved time series forecasting of healthcare consumption and life trajectories of Medicare beneficiaries
    6. Created and informed new methods - in conjunction with external consulting firm - to more accurately forecast future demand for 988 crisis service utilization. I leveraged both external data sources and Medicaid inpatient claims - in order to inform Congressional funding and to expedite collaboration between CMS and SAMSHA
  • Supervisors: Misu Tasnim and Maya Israni
• July 2021 - November 2021: Senior Data Scientist
  • Cisco Systems Incorporated - Sales.AI Group
  • Achievements:
    1. Developed quantile regression model for estimating sales pipeline revenue - piloted in Asia-Pacific Sales region
    2. Improved forecasting models and forecasting diagnostic methods for future service opportunities in the sales pipeline - expanding 300 million dollars in recurring revenue
    3. Proposed a novel prototype product and services recommendation service using probabilistic matrix factorization
    4. Supervised and mentored a team of junior data scientists spanning Singapore, Vancouver, and San Jose
    5. Served as a volunteer data scientist with Code for America on the GetCalFresh project - developed elementary constituency parsing and trigram model for customer service requests
    6. Demand-pricing optimality using Thompson Sampling and Bayesian inference
  • Supervisors: Mr. Kunal Shukla and Mr. Seokho Hong
• January 2020 - July 2021: Data Scientist
  • Cisco Systems Incorporated - Advanced Security Initiatives Group
  • Achievements:
    1. Co-authored Patent on Endpoint Monitoring using (Please see item 5, below)
    2. Continued developing Zero-Trust solutions against privileged insider threats - primarily using Fourier analysis and methods derived from signal processing and estimation theory
    3. Applying spectral analysis and developing Bayesian estimators for change-point detection in order to characterize anomalies in device telemetry communications from around the world. The use case is for detecting counterfeit devices.
    4. Regularization methods for non-parametric inference to estimate frequency and severity of release-on-release incidents
    5. Deep Graph Neural Networks for link prediction and node classification within the context of network architectures for products and services.
    6. Consulting on and assessing NMT language models - across various languages - for PII that may affect legal compliance.
  • Supervisor: Mr. Joseph Gipson
• May 2019 - December 2019: Data Scientist
  • Cisco Systems Incorporated - Security and Trust Organization
  • Achievements:
    1. Zero-Trust: Developing signal processing algorithms for detecting privileged insider threats
    2. Spearheaded new statistical learning approach for identifying counterfeit in telemetry data
    3. Submitted new patent for anomaly detection in defect-tracking systems for development artifacts
    4. Implemented POC novel deep learning bi-LSTM archiecture for search and classification of forensic evidence
    5. Created ARIMA, SARIMA and AR models for extreme-event forecasting of revenue affects after public disclosures
    6. Applied survival analysis estimation for user engagement studies for compliance training
    7. Presented work to executive leadership and liased with stakeholders for continual feedback
  • Supervisor: Mr. Jeff Schutt
• August 2019 - March 2020: Research Assistant
  • North Carolina State University
  • Achievements:
    1. Developed online-training and batch experimentation architecture for transfer learning applied to autonomous driving.
    2. Worked on theoretical derivation of function approximation derivation for off-policy exploration
  • Supervisor: Dr. Eric Laber, Department of Statistics
• January 2019 - May 2019: Research Assistant
  • North Carolina State University
  • Achievements:
    1. Implemented Asynchronous Actor Critic Models (A3C) for NFL football and NBA basketball
    2. Formulated and published research report comparing approaches - including Dueling DQN Prioritized Replay - within the context of first-player shooter games
    3. Researched, replicated and extended MDP and Finite Horizon Reinforcement Learning models for spatial decision making
  • Supervisor: Dr. Eric Laber, Department of Statistics
• May 2018 - August 2018, Automation Engineering Intern
  • Compozed Labs, Allstate
  • Achievements:
    1. Created novel RNN speech generation language model to be integrated into RoadSide Service
    2. Architect and implement a proof-of-concept event-driven microservice, written in Golang, that leverages RabbitMQ as a broker and Redis as a persistent backend. Built additional configuration for client to also accommodate Amazon SQS and Memcache as a job queue.
    3. Developed custom API schema validator package in order to prevent invalid response and requests to be received from third-party vendor applications. The impetus for this features was to adapt to the lack of a consumer-driven contract implementation.
    4. Created and presented a workshop on advanced functional Java concepts for machine learning
• December 2017 - October 2018: Research Assistant
  • North Carolina State University
  • Achievements:
    1. Architected, built, tested and deployed application that visualized predictions from an online deep neural network classifier trained on random Voronoi partitions of a spatial domain. This deep neural network yields remarkably accurate geolocation predictions. When applied to the microbiomes of over 1,300 dust samples collected across the U.S., more than half of predictions produced by this model fall within 90 kilometers of their origin, a 60% reduction in error from competing geolocation methods.
  • Supervisor: Dr. Eric Laber, Department of Statistics
• August 2017 - December 2017: Research Assistant
  • North Carolina State University
  • Achievements:
    1. Researched and implemented ad Gated Feedback Recurrent Neural Network that predicts the likelihood that a escort post is from a trafficking victim from the raw text.
  • Supervisor: Dr. Eric Laber, Department of Statistics
• December 2016 - July 2017: Software Engineer
  • Social Tables
  • Achievements:
    1. Cofounding member of the Infrastructure Team building service architecture, data engineering tools and data securitization middleware. This included rewriting roles and permissions management for all applications and building a core data platform to provision all applications with proprietary customer data.
    2. Rearchitected entire front-end of legacy product into mobile-first application – delivered $40,000 in revenue within first two weeks of re-release: Event Sales Solution
• April 2016 - November 2016: Software Engineer
  • DivvyCloud
  • Achievements:
    1. Spearheaded the development of an automated testing, orchestration and deployment framework using Docker, Python and AWS S3
    2. Prototyped and deployed memory and request monitoring data pipeline for production deployments using Elasticsearch, Akka, and Scala. Result: Reduced false positive rate for customer-reported memory leaks and excessive resource utilization by 72% in one quarter
• June 2015 - March 2016: Associate Technologist
  • Wynyard Group
  • Achievements:
    1. Developed and optimized ETL pipelines using Cassandra, PySpark, Python, Pandas and Spark in order to deploy logistic regression model for classifying anomalous financial transactions.
    2. Applied and tested n-gram models, tf-idf, Hidden Markov Model, Bayesian Bag of Words, in order to describe and to classify audio transcripts and forensic evidence sourced from law-enforcement and national security agencies.
    3. Developed visualizations using d3.js, AngularJS and crossfilter.js, in order to present exploratory data analysis and data mining efforts for geospatial classification of entities for law enforcement agencies.

Skills

• Fluency in statistical learning theory and application using (among others):
  • Pytorch, Tensorflow and Keras
  • SciPy, scikit-learn, statsmodels, and lifelines
  • PyMC3 an ArviZ
  • Numba
• Natural Language Processing with Deep Learning
  • Named Entity Recognition and constituency parsing
  • Word window classification
  • Dependency parsing
  • Contextual Word Representations
  • Seq2Seq and attention networks
• Distributed Computing and Data Engineering leveraging:
  • Scala
  • Airflow
  • Kafka
  • Spark and Spark ML
  • Snowflake
  • Google Compute Engine
  • BigQuery
  • Apache Arrow
• Bayesian Estimation and Hierarchical Models:
  • Latent-Dirichelet Allocation
  • Non-parametric hypothesis testing (two sample, etc.)
  • Clustering with Gaussian Mixtures
• Causal Inference Theory and Application including:
  • Propensity Scoring
  • Matching
  • Difference-in-Difference estimation
  • Mixed Effects Models for hierarchical models

Publications

Multi-angled Statistical Approach to Human Trafficking Detection and Profiling

Saanchi, Y, Wang, M., Ahluwalia, S., Laber, E. & Caltagirone, S. (2017, December). Multi-angled Statistical Approach to Human Trafficking Detection and Profiling. Laboratory for Analytical Sciences, Raleigh, NC.

Repucall: Towards Describing the Reputability of Phonenumbers

Ahluwalia, S. & Reaves, B. (2018, April). Repucall. Paper submission at the North Carolina State University OUR Symposium, Raleigh, NC.

Deep Learning within the Context of Doom

Ahluwalia, Saran and Laber, Eric. (2019, May). " Deep Learning within the Context of Doom" Unpublished Manucript

Review of Joint Clustering with Correlated Variables

Ahluwalia, Saran, Kim, Minjee and Park, Hannah. (2020, May). " Review of Joint Clustering with Correlated Variables" Unpublished Manucript

Estimating Single-Family Returns and Home Price Appreciation

Ahluwalia, Saran. " SFR Returns and HPA - A cross-section study across the United States" Working Paper

Talks

Predicting Gentrification in District of Columbia Wards

October 17, 2015

Symposium Talk at District Data Labs Fellowship Symposium, Washington, DC

Multi-Angled Statistical Approach to Human Trafficking Detection & Profiling

December 08, 2017

Conference proceedings talk at 2017 LAS Research Symposium, Raleigh, NC

Enabling Search Using bi-LSTMs

August 07, 2019

Talk at Cisco Systems International Data Science Symposium, San Francisco, California

Teaching

Capital Teaching Resident

AP Biology Teacher

Upper School Educational Technologist