Pratyush Narain

Software Engineer

Bengaluru, Karnataka, India4 yrs 11 mos experience

Key Highlights

Proven expertise in low-latency systems and network protocols.
Achieved significant memory optimization in load balancing.
Led strategic initiatives impacting over 900 sites.

Stackforce AI infers this person is a Networking and Software Development specialist with a focus on performance optimization.

Contact

Skills

Core Skills

NetworksData Plane Development Kit (dpdk)DebuggingOptimizationTraffic ManagementPythonTraffic Analysis

Other Skills

NUMALoad BalancingCPU ProfilingCongestion ManagementWiresharkTraffic ClassificationData AnalysisSoftware DevelopmentJavaDocker SwarmPostgreSQLGitSystem DesignLow Level ProgrammingLinux

About

Padho With Pratyush https://www.youtube.com/@padho_with_pratyush I'm a Software Engineer 2 with 2+ years of experience, currently driving performance and innovation at Akamai. I specialise in system-level programming, network intelligence, and high-performance software development using C, C++, and Python. A proud graduate of IIT Dhanbad, I bring strong fundamentals in data structures, algorithms, and software design. At Akamai, I am working in the Cloud Data Services team where I am responsible for handling low latency network logs for billing and reporting. 🔍 What I Bring: Hands-on experience in network protocols, multithreading, and low-latency systems Strong expertise in debugging, optimization, and performance tuning Proven ability to design and implement robust, production-grade features Experience with Linux internals, version control (Git and Perforce), and Agile workflows

Experience

4 yrs 11 mos

Total Experience

2 yrs

Average Tenure

9 mos

Current Experience

Akamai technologies

Software engineer 2

Aug 2025 – Present · 9 mos · Bangalore

Sandvine

2 roles

Software Engineer 2 @ Sandvine

Promoted

Jun 2023 – Aug 2025 · 2 yrs 2 mos

Numa Aware Load Balancing
Implemented NUMA Affine Load Balancing in a 1/2/4 Numa System for IPV4, IPV6 and GRE/GTPU Tunnel Traffic
Goal of the feature was to process all the subscriber traffic in a single numa node in a multi-socket system and scale it to over 4 Million
Subscribers
Resulted in a 50% improvement on memory consumption due to localisation of shaping objects created per NUMA
Used concepts of bit masking on IP addresses using 32 and 128 bit spec and masks to create flow rules on 100G mellanox NICs
CPU Profiling:
Feature Lead of implementing a trigger based CPU profiler on the engine of the
flagship product Active Logic.
Came up with an approach to store the register addresses and translating the top 10
addresses in human-readable symbols
Resulted in a {50%} reduction in turn around time for debugging packet drops
Intent Based Congestion Management
Increased network efficiency by {15%} by implementing congestion management and
intelligent traffic shaping solution
Spearheaded strategic initiatives for deploying a traffic management solution impacting
more than 900 sites in NOS Portugal

NetworksData Plane Development Kit (DPDK)

SDE-1@ Sandvine

Jun 2023 – Dec 2023 · 6 mos

# Worked as a part of Content Classification. The feature aimed to classify internet traffic into applications and further category wise such as file downloads, gaming etc. The feature added to the product as a big improvement in App-Classification. Worked on python, Jupyter Notebook and Wireshark to analyze traffic patterns, differentiating between TCP/UDP and associating patterns to categories.

PythonWiresharkTraffic Analysis