Harmandeep Singh Email & Phone Number

Bengaluru, Karnataka

Bengaluru Area, India

Gurgaon, India

- NetSim - Network simulator for simulating any network based system. It allows you to predict/simulate the "what if" scenarios in the Delhivery network. Allowing us to answer questions about network impact without making the actual ground level changes. - Developing a mathematical model based on Decision tree regression for predicting the time a shipment will spend in a facility, or in a truck if being transferred from one facility to another. - Based on a similar model for predicting the next facility the shipment will go to, given it's final destination - Developing a Queue-based API service architecture to simulate millions of packages. - Tech Stack - python3, OpenScoring server, PySpark MLlib, AWS EMR - Polygon - An open source library written in C++, with python bindings, for use in various projects in Delhivery for making convex/concave hulls. - Implemented convex hull algorithm based on Graham's scan method. - Implemented the concave hull algorithm from a research paper which extended the mentioned convex hull algorithm. - Tech Stack - C++, boost-python - Mapper - Locality-Facility management tool for system flexibility and optimization. It maintains the mapping of locality and it's serving facility, thus, allows the users to maintain a dynamic system which in turn helps in moderating the load and distance each facility ends up serving. - Implemented a distance and load based system for equal distribution of work among the facilities in a city. - Developed an async job queue for running jobs related to remapping the localities of a city. - Developed an algorithm for automatically verifying the new mapping based on point in a polygon logic. - Tech Stack - Django, Celery, PostgreSQL, ElasticSearch, RabbitMQ

Gurgaon, India

- Farzi - A fast fuzzy matching system that allows us matching two strings. It's based on Edit Distance but with dynamic costs, which are algorithmically determined based on mistakes that people make. - Created a dynamic system for finding the replacement costs between various pairs. - Implemented a lookback system, which allows replacement of multiple characters with one or multiple characters. Eg. - 'f' and 'ph'. - Implemented multi level hashes for fast access of pairwise costs. - Used heuristics for fast termination of negative cases. - Tech Stack - C++, boost-python, gperf - ASE - A heirarchy based algorithm for clustering similar addresses into groups. Used for recognizing same/duplicate sellers accross multiple clients. - A simple Disjoint set based implementaton for maintaining the groups. - Subsequence level comparison between the heirarchy of two addresses for determining the similarity. - Tech stack - python3 - Report Central - Data visualisation module for internal analytics at Delhivery that acts as a charting service for different level of operations within the organisation. - Tech stack: AngularJS, Node.js, ElasticSearch, Redis

Noida Area, India

- AT&T addressbook - Application that allows phone's contacts and groups to be synced with a server. - Implemented the contact image sync by converting the jpg to base64 format. - Improved the sync sanity process which reduced the boot time of phone by 3-4 seconds. - Developed the AT&T proprietary extension for syncing the groups in the phonebook. - Went to South Korea, for getting the design reviewed from a third party, WebSync co. ltd. - Tech stack - C, SyncML, WBXML - Samsung Kies - Client(Phone) side handling of Samsung Kies.

Bachelor Of Engineering (B.E.), Computer Engineering

Education record

Education record