Vaibhav Singhal

Vaibhav Singhal

vaibhav@portfolio ~ %
_

I build backend systems that serve millions — identity platforms at Workday, voice AI at LivePerson, serverless pipelines at Black Knight. I care about resilient architecture, clean APIs, and making other engineers faster.

0+ Years Experience
0+ Systems Built
0 Certifications
0+ Certificates
Email LinkedIn GitHub Contact Resume
Scroll to explore
01

Work Experience

Building distributed systems at scale across identity, voice, and fintech.

Senior Software Engineer

Workday (Evisort)

IdentitySCIMOIDCMulti-tenantCPS

Owning the identity platform post-acquisition — SCIM provisioning, unified OIDC auth, single logout, per-tenant schema isolation, and a resilient CPS publish queue with exponential backoff retries. Serving thousands of auth requests/min across Workday and Evisort.

Oct 2024 – Present Toronto, ON

◆ Architecture & System Design

Built SCIM API endpoints for Workday-Evisort user provisioning
  • Delivered the first Workday-Evisort integration post-acquisition (CRUD + Restore) in the identity service.
  • Implemented dynamic model creation for provisioning flexibility across Workday and the legacy identity provider.
  • Led GA hardening: deterministic role mapping ("highest role wins"), SCIM import failure fixes, and Workday identifier handling for migration stability.
Improved identity and document platform services for Workday user provisioning
  • Established a dedicated internal service-to-service channel (HTTP Signature Auth, dedicated port) for secure provisioning.
  • Fixed 2 critical IDOR vulnerabilities in document version endpoints — changed cross-tenant access from 403 to 404.
  • Built an async user creation task for post-SCIM user provisioning with retry logic and outbox events.
Implemented unified authentication via OIDC
  • Built a high-throughput token exchange endpoint handling tens of thousands of requests per minute.
  • Designed OIDC callback handling across identity and auth services (JWT generation, token verification, identifier resolution).
  • Integrated cookie-based session management and state-based redirection for deep linking.
Designed end-to-end unified logout between Evisort and Workday
  • Implemented frontchannel (Evisort-initiated) and backchannel (Workday-initiated) Single Logout flows.
  • Built dynamic JWT logout token validation using internal certificate authority discovery.
  • Demoed the complete logout flow to the engineering team and company-wide.
Owned the Super Admin Toggles EPIC for migration-safe identity architecture
  • Drove full technical design, ticket decomposition, and implementation across 4 services and the UI.
  • Introduced per-client toggles (user_management, authentication_provider) to decouple tenant mapping from login enforcement.
  • Designed backward-compatible data migration with feature flag controls for safe rollout.
Architecting per-client PostgreSQL schema isolation for the auth gateway
  • Replacing shared public schema with per-tenant isolation to reduce cross-tenant blast radius.
  • Implementing dual-write architecture with feature-flagged reads for zero-downtime migration.
Built CPS Publish Queue with retries and exponential backoff
  • Designed a DB-backed task queue so all CPS publish operations (REGISTER, DEREGISTER, UPDATE, RECREATE) are enqueued asynchronously and processed by a dedicated worker.
  • Implemented exponential backoff retries (base 30s, max 600s, 6 retries) — CPS publishes survive 12+ hours of downtime during quarterly hybrid patching.
  • Built atomic RECREATE (deregister old → register new) with partial-failure recovery and snapshot-based DEREGISTER resilience.
  • Reduced Sentry alerts from ~50 per task to max 6; created a datapatch to re-enqueue previously failed tasks.

◆ Technical Leadership

Conducted live engineering demos and led design reviews
  • Presented SCIM, UPW, and unified authentication at Evisort Tech Demo and Eng Syncs.
  • Documented workflows and recorded sessions to support team onboarding.
Owned on-call rotations and incident response
  • Reviewed and documented 5+ past incidents (SCIM sync failures, token verification issues).
  • Built a Confluence debugging playbook and mapped service ownership across 4+ platform services.
Buddied 2 new hires; participated in 14 interview loops (3 successful hires)
  • Created a repeatable onboarding reference guide.
  • Provided feedback to improve onboarding documentation and materials.
Led Operational Readiness Reviews (ORR) for the auth gateway
  • Authored ORR documentation aligned with Workday reliability standards.
  • Enabled Prometheus-based observability and alerting; extended instrumentation to the NestJS gateway.
  • Raised authentication test coverage to 80%+.

◆ Development & Operations

Improved platform stability and scalability across identity and document platform services
  • Resolved key integration bugs and profiled the high-traffic token exchange endpoint.
  • Refactored core components for maintainability and increased test coverage.
Drove cross-service integration resilience
  • Built robust SSO token validation using internal CA certificate discovery.
  • Enabled auth gateway observability via /ping and /health endpoints.
  • Standardized metadata propagation across platform services and UI.
  • Refactored auth library naming conventions across Python and TypeScript codebases.

Software Engineer

LivePerson Inc

Voice AIKafkaMicroservicesGCP

Built Voice AI platform end-to-end — media microservices, SIP APIs, and Kafka schema service. Cut latency 35% and scaled concurrent calls to 700.

Jan 2022 – May 2024 NYC → Toronto

◆ Architecture & System Design

Rebuilt Connector Service — Reduced data event size by 40% and latency by 35%.
  • Identified bottlenecks in inter-service communication.
  • Designed an optimized event pipeline with Redis-backed brand configuration lookups.
Redesigned Media Service into microservices — Increased concurrent calls by 50% (450 to 700).
  • Decomposed monolithic service into TTS, STT, and media microservices.
  • Defined and implemented the observability strategy (metrics, monitoring, alerting).
Owned Audio File Management Service end-to-end — Drove 25% increase in user engagement.
  • Architected storage and retrieval layer for audio blobs and Azure client keys using GCS and Redis.
  • Integrated Azure Text-to-Speech for multi-language support via SSML.
Proposed and drove adoption of Centralized Schema Service — Reduced integration overhead by 40% and debugging time by 25%.
  • Identified fragmented Kafka schema management as a cross-team pain point.
  • Designed the service to enforce schema evolution and consistency.
  • Authored internal documentation and onboarding guides adopted across teams.
Designed and delivered SIP Connection APIs — Increased customer retention by 20%.
  • Built SIP connection support (IP, FQDN, Credentials) using Telnyx SDKs, Java, and Micronaut with Kafka.

◆ Development & Operations

Optimized shared utility libraries — Improved code efficiency by 22% across internal services.
  • Tools: Redis, Log4j, GCS, Firestore.
Built automated OOM diagnostics — Reduced debugging time by 25%.
  • Automated thread and heap dump backups to GCP buckets during Out-of-Memory incidents.
Secured transaction logs and reduced on-call burden — Cut on-call debugging time by 4 hours weekly.
  • Built bash tooling for log retrieval from Kubernetes pods and GCP Logs Explorer.
Drove data security improvements
  • Implemented data redaction for NPIs, auth tokens, and credentials across GCP logs and Kubernetes pods.
  • Eliminated sensitive data exposure in production logs, enabling compliance with enterprise data handling requirements.
Raised test coverage from 36% to 87% — Cut testing costs by 20%.
  • Used JUnit and JMeter; established testing standards for the team.
Led Inbound Call Automation Bots initiative
  • Established a fully operational system for weekly performance runs.
  • Built the initial proof of concept and coordinated cross-team end-to-end testing.
  • Delivered knowledge transfer to ensure long-term ownership.

Software Engineer

Black Knight Inc

AWSOCRPythonServerless

Architected serverless APIs on AWS — improved OCR accuracy to 89%, cut production costs 20%, and automated 80% of manual transaction retries.

Jun 2020 – Jan 2022 Philadelphia, PA

Architected and developed APIs for Input & Output Handler Services on the AIVA (Artificial Intelligence Virtual Assistant) team.

Built customer environment management APIs — Reduced production costs by 20%.
  • Utilized AWS services (API Gateway, Lambda, S3, DynamoDB, SSM).
Designed OCR data extraction microservices
  • Increased data accuracy from 80% to 89% and improved performance by 25%.
Led greenfield service development
  • Built resilient services and orchestrated infrastructure stacks using Python, CloudFormation, and S3.
Resolved provisioning and build pipeline failures
  • Reduced testing load times and lowered AWS costs across environments.
Automated transaction anomaly recovery using Python and Boto3.
  • Eliminated 80% of manual retry attempts caused by code defects or AWS outages.
Earlier Experience

Research Scholar

Bharati Vidyapeeth's College of Engineering

SteganographyNLPCloud Security

Published two research papers on cloud security steganography and NLP-based clinical decision support systems.

Jan 2017 – Sep 2017 Delhi, India

CSE Department under Dr. Rachna Jain — Published two research papers.

Enhance Data Security in Cloud Using Steganography
  • Studied Cloud Service Providers vs Internet Service Providers based on cost, speed, scalability, and performance.
  • Reviewed techniques of steganography (LSB) and digital watermarking (DFT) in spatial and frequency domain.
  • In-depth study on Symmetric Key Encryption (DES, AES) and Asymmetric Key Encryption (RSA).
  • Proposed a model for Image Steganography using LSB, mapping every bit of data and encryption key in RGB pixels.
Medical Application Using NLP Over Cloud
  • Knowledge acquisition on NLP systems and open source libraries (NLTK, Polyglot, Pattern).
  • Studied Morphological Processing, Syntax, Semantic and Pragmatic Analysis.
  • Comparative study on public, private, and hybrid clouds based on reliability, cost, and scalability.
  • Proposed a computerized Clinical Decision Support (CDS) System using cloud services.

Consulting & Software Intern

TATA Computer Management Corp

PythonFlaskMongoDBETL

Built a full-stack web app with Flask and MongoDB — drove 45% sales increase and 30% faster query performance via ETL optimization.

May 2016 – Aug 2016 Delhi, India
Full stack API-driven web application — Increased sales by 45% and reduced data storage inventory by 38%.
  • Implemented Agile process across teams, improved customer interaction and sprint feedback.
  • Developed RESTful APIs using Python with Flask, JWT Session Management for security.
  • Performed ETL on legacy database, improved query performance by 30%.
  • Interacted with the DB using MongoDB Alchemy (NoSQL), improving flexibility and providing Object-Oriented structure.

Android App Development Trainee

Bharati Vidyapeeth's College of Engineering

AndroidJava

Developed an Android video player with gesture controls, custom UI, and third-party SDK integrations.

May 2015 – Jul 2015 Delhi, India
Developed a video player application using Android Studio.
  • Built list/grid view and gesture controls.
  • Worked on app UI, deep functionality, and third-party SDKs.
02

Technical Arsenal

Technologies I work with daily and at depth.

CORE
Python
Java
TypeScript
Bash
C/C++
SSML
FastAPI
Micronaut
NestJS
Flask
Django
Kafka
RabbitMQ
OIDC
SCIM
JWT
Docker
PostgreSQL
Redis
MySQL
MongoDB
Firestore
DynamoDB
S3
GKE
Kubernetes
CloudFormation
API Gateway
Compute Engine
Prometheus
Grafana
Sentry
GitLab CI/CD
JMeter
Swagger
Gradle
Spark
Hadoop
Sqoop
Jira

Languages

PythonJavaTypeScriptBashC/C++SSML

Database & Storage

PostgreSQLRedisMySQLMongoDBFirestoreDynamoDBCloud StorageS3

Back-end

FastAPIMicronautNestJSFlaskDjangoKafkaRabbitMQOIDCSCIMJWTDocker

Cloud (GCP & AWS)

GKEKubernetesCompute EngineLoad BalancingCloud BuildCloudFormationAPI Gateway

Big Data

Spark StreamingHadoop HDFSSpark RDDDataFramesSqoopApache Flume

Tools & Observability

GitLab CI/CDPrometheusGrafanaSentryJMeterSwaggerGradleJira
03

Education

2018 – 2020

Arizona State University

Tempe, Arizona, USA

M.S. Software Engineering

2014 – 2017

Guru Gobind Singh Indraprastha University

Delhi, India

B.Tech Computer Science

2011 – 2014

Guru Nanak Dev Institute of Technology

Delhi, India

Associate — Electronics & Communications

04

Certifications

CCA-175

Cloudera Certified Associate

Spark and Hadoop Developer

PCAP

PythonInstitute Certified Associate

Python Programming (PCAP-31-02)

OCA

Oracle Certified Associate

Database SQL (1Z0-071)

GCP

Google Associate Cloud Engineer

Google Cloud Platform

Data Warehouse — The Ultimate GuideSoftware Architecture & Design of Modern Large Scale SystemsMicronaut — Cloud Native Microservices with JavaGCP Associate Cloud Engineer — Google Cloud CertificationSpark and Hadoop Developer — Python (PySpark)Spark and Python for Big Data with PySparkREST APIs with Flask and PythonAdvanced REST APIs with Flask and PythonGCP: Complete Google Data Engineer & Cloud Architect GuideOracle SQL Developer From Scratch
05

Projects

Hover to flip and see details.

Ice Hockey

RosterData — Ice Hockey v2

PythonScrapyPostgreSQLAWS

RosterData — Ice Hockey Statistics v2

Jan 2020 – Apr 2020

Expanded RosterData from NHL-only to 4-league coverage (NHL, SHL, Liiga, KHL) with cross-league player comparison.

  • Built league-specific Scrapy pipelines for SHL, Liiga, and KHL — each with unique HTML structures and pagination patterns.
  • Designed a normalized JSON data model for cross-league consistency (player identity, season stats, team associations).
  • Stored normalized data in PostgreSQL on AWS RDS with query-optimized indexing for player lookup and league standings.
  • Built REST APIs for player search, team rosters, and cross-league career timelines.
TF-IDF

LSA Classification & Prediction

Pythonscikit-learnFlaskGCP

Latent Semantic Analysis — Classification & Prediction

Mar 2020 – Apr 2020

End-to-end ML pipeline for financial document classification — 98% accuracy, deployed as a REST API on GCP.

  • Preprocessed financial documents: tokenization, stop-word removal, TF-IDF vectorization into numerical feature vectors.
  • Applied LSA (SVD on TF-IDF matrix) for dimensionality reduction while preserving discriminative dimensions.
  • Evaluated Naive Bayes, SVM, Random Forest, and Logistic Regression with grid search and cross-validation.
  • Built Flask REST APIs for real-time inference and deployed containerized on GCP Compute Engine.
Scikit-Learn

Diabetes Classifier & Clustering

PythonPandasNumPyscikit-learn

Diabetes Patient Classifier & Meal Clustering

Aug 2019 – Nov 2019

Rigorous ML classification on clinical data with K-fold validation — 85% accuracy, plus unsupervised meal clustering.

  • Preprocessed Pima Indians Diabetes Dataset: median imputation for missing values, feature normalization, class imbalance analysis.
  • Evaluated Decision Tree, SVM, and KNN classifiers using K-fold cross-validation for generalizability.
  • Applied K-Means and DBSCAN clustering on meal/glucose data to identify dietary patterns correlated with glucose response.
  • Achieved 85% accuracy / 83% confidence validated across folds — not an artifact of a favorable split.
Math Portal

Visual Learning Portal

PythonFlaskSQL-AlchemyAgile

Online Visual Learning Portal

Aug 2019 – Nov 2019

Interactive math portal with drag-and-drop, built by a team of 5 using Agile/Scrum and formal design patterns.

  • Applied Facade (simplified complex subsystems), Factory (dynamic content creation), and Iterator (collection traversal) patterns.
  • Built Flask REST APIs with SQLAlchemy ORM; drag-and-drop frontend for interactive math exploration.
  • Ran 2-week sprints with backlog grooming, standups, and retrospectives across a 5-person team.
Programming Language

!Xobile Programming Language

PythonPrologCompiler Design

!Xobile Programming Language

Jan 2019 – May 2019

Custom OOP language with full compilation pipeline — grammar spec, tokenizer, parser, and semantic analyzer.

  • Designed formal grammar: class declarations, inheritance, control flow, expressions, and OOP constructs (this, constructors, method dispatch).
  • Built regex-based tokenizer in Python (keywords, identifiers, operators, string literals with escape chars).
  • Implemented parser and semantic analyzer in Prolog: type checking, variable scoping, inheritance cycle detection, and method resolution.
  • End-to-end pipeline: valid programs produce semantically validated parse trees; invalid programs yield meaningful error messages.
Hadoop

Hadoop Cluster & MapReduce

HadoopMapReducePythonHDFS

Hadoop Cluster & MapReduce API

Mar 2017 – Jun 2017

8-node Hadoop cluster processing 300M+ YouTube video logs with MapReduce and cross-version benchmarking.

  • Configured 1 NameNode + 7 DataNodes with HDFS replication and YARN resource management.
  • Designed MapReduce jobs (Hadoop Streaming, Python): most-viewed by category, upload trends, viral video detection.
  • Benchmarked identical jobs across Hadoop versions — measured job time, CPU/memory, shuffle overhead, and speculative execution.
  • Documented quantitative speedups in scheduling and data shuffle across framework versions.
06

Let's Connect

Open to conversations about distributed systems, identity platforms, and interesting engineering challenges.

LinkedInLinkedIn
GitHubGitHub
EmailEmail
Contact
Resume