Steve Biju John.
Steve Biju John.

EXPLORE / STEVE BIJU JOHN

Mechanical engineering & intelligent systems

Thoughtful design.
Practical engineering.
Ideas brought to life.

I’m Steve Biju John, a mechanical engineer exploring robotics and intelligent systems, with a creative background in visual design.

Find out moreSelected projects

Find me on GitHub and LinkedIn, or say hello.

Mechanical designRobotics & mechatronicsSimulation & fabricationVisual communication

A little about me

01 / ABOUT

I enjoy the space where engineering and creativity meet.

My background is in mechanical engineering, with hands-on experience in design, fabrication, testing, and manufacturing. My projects range from a six-axis desktop robot arm to a laser CNC and a system for monitoring dining-hall occupancy.

I’ve also worked in graphic design and multimedia, creating brand identities, product catalogs, and video content. That experience shapes how I communicate technical ideas: clearly, visually, and with the audience in mind.

In September 2024, I began master’s studies in Embedded and Intelligent Systems at Halmstad University in Sweden.

Selected projects

02 / WORK
Latest project · Master’s thesis

DRIFT-IFT-GNN

Information flow tracking for hardware Trojan detection.

Hardware securityInformation flow trackingGraph neural networks
View thesis project on GitHub

My thesis project uses two complementary representations of hardware designs to prepare training data for graph neural networks (GNNs) that detect hardware Trojans.

Gate-level analysis

Tracks information flow through a synthesized data-flow graph (DFG).

Source-level analysis

Examines the Verilog abstract syntax tree (AST), preserving logic that synthesis may remove.

Robotics / 01

Six-axis desktop robot arm

A 3D-printed robot arm designed and fabricated for basic pick-and-place operations.

6 DOF3D printingMotor control
Inside the project

I designed and fabricated the arm and developed software to control the working range of each stepper motor.

Mechatronics / 02

Project Rush Hour

A dining-hall occupancy system to help students find seats during the rush between classes.

MechatronicsPeople countingOccupancy display
Inside the project

The system used mechatronic controllers to count and display the number of people inside the hall, helping students decide where to go during peak lunch hours.

Fabrication / 03

Laser CNC

A laser engraving machine built to work with non-reflective surfaces, including wood-like materials and suitable plastics.

CNCG-codeFabrication
Inside the project

Building the machine expanded my understanding of CNC operation, G-code, and manufacturing applications.

Simulation / 04

Backward-facing step flow

A computational fluid dynamics study exploring how flow behaves over a backward-facing step.

ANSYS FluentCFDFlow analysis
Inside the project

I created the backward-facing step geometry and studied the flow over its surface using ANSYS Fluent.

Where I’ve worked

03 / EXPERIENCE
Feb – Aug 2023Dubai, UAE

Sentient Labs FZ LLC

Mechanical Design Engineering Intern

Worked on the development and fabrication of a winch system for unmanned surface vehicles, along with 3D-printed structures supporting GPS antenna performance.

Contributed to engineering proposals, product testing, and the PLECO product line’s brand guidelines and marketing materials.

Jun – Aug 2022Chennai, India

Hyundai Motors India Ltd

Engineering Intern

Researched engine-shop reliability through failure mode and effects analysis (FMEA), and evaluated safety checkpoints across machinery and assembly sections.

Nov 2020Dubai, UAE

Sealmech Trading LLC

Engineering Trainee

Helped resolve customer engineering challenges and improve inventory organization, reducing component search time from around five minutes to under two minutes.

Do not Stop personal poster

Personal posters

Typography, texture & visual experiments

  • Poster experiments

    Personal explorations in expressive typography, image-making, and composition.

View personal posters
TEDx Bennett University speaker poster for Usha Uthup

TEDx Bennett University

Event graphics & speaker posters

  • Incognito speaker series

    Speaker posters for the TEDx Bennett University event.

  • Event communication

    Brochure design and social media video.

View TEDx collection
PLECO LinkedIn banner: Water Management Redefined

Sentient Labs

PLECO brand identity & product communication

  • PLECO brand system

    A product-family identity map connecting the PLECO ecosystem.

  • Product communication

    Brochure, product-video, LinkedIn banner, and website-design collections.

View Sentient Labs collection
Project Qadira logo and visual identity

Project Qadira

Brand building & visual identity

  • A brand with a purpose

    Collaborative branding for an Enactus initiative supporting accessible menstrual care.

  • Identity & illustration

    Logo design and visual storytelling communicating the project’s core beliefs.

View Project Qadira on Behance

More design collections

Moshi MoshiGrasp MediaSealmechFreelanceBURS

Learning & leadership

05 / BACKGROUND
Sep 2024 – ongoing

Master’s in Embedded & Intelligent Systems

Halmstad University
Halmstad, Sweden

Graduated 2023

B.Tech in Mechanical Engineering

Bennett University
Greater Noida, India

Enactus Multimedia Head

Nov 2020 – Jun 2021

Led a team of 13 creating visual communications and brand-aligned deliverables.

Graduated 2019

IB Diploma Programme

Kohinoor American School
Khandala, India

Student Body President

Aug 2017 – Aug 2018

Led the student council, addressed student concerns, and supported cultural and academic activities.

My toolkit

06 / SKILLS

Design & manufacturing

SolidWorks, Fusion 360, CATIA V5, AutoCAD, additive manufacturing, CNC

Code & simulation

Python, C, ROS, MATLAB, ANSYS, UiPath, Ubuntu Linux

Creative tools

Blender 3D, Adobe Creative Suite, Microsoft Office

Languages
English
Fluent / native
Malayalam
Fluent / native
Hindi
Fluent / native
Swedish
Basic proficiency
Arabic
Basic proficiency

Let’s connect

Have something in mind?
I’d like to hear about it.

Get in touch about engineering, robotics, or visual design.

Email me
GitHubLinkedInstevecanwork247@gmail.com
DRIFT / Master’s thesis ↗

HARDWARE SECURITY / INFORMATION FLOW / GRAPH LEARNING

DRIFT-IFT-GNNTwo views of hardware.
One richer training signal.

A dual-representation RTL information flow tracker for hardware Trojan and side-channel detection.

Why two representations?

DRIFT traces how secret inputs, such as keys and state registers, influence signals in Verilog hardware designs. It combines a synthesized data-flow graph (DFG) with the original source’s abstract syntax tree (AST), then fuses their scores into labels for training graph neural networks.

Synthesis can remove logic with no digital output. A power-side-channel Trojan may still use that logic to leak information through power consumption. The AST path preserves source-level evidence that the gate-level path can lose.

Gate-level / DFG

Yosys builds the netlist. Information flow propagates through circuit connections to capture digital signal paths.

Source-level / AST

PyVerilog analyzes the raw source, including declared logic that synthesis may discard. The guide notes that this path can miss cross-module port connections.

DRIFT sample scatter plot comparing DFG and AST scores, highlighting signals detected by the AST path but missed by the synthesized DFG.
What the second view reveals. Points above the diagonal have stronger AST scores. The repository highlights AES-T2100’s COUNTER, Tj_Trig, and LEAKBit as examples of source-level evidence lost in synthesis. Original GitHub figure ↗

From RTL to training data

  1. 01

    Build both representations

    Synthesize the Verilog netlist and extract the source AST.

  2. 02

    Track and score information flow

    Propagate taint from sensitive inputs, then apply golden-delta, GLRA, and QtFlow scoring.

  3. 03

    Fuse signal-level labels

    Combine DFG and AST evidence, recording agreement and cases detected by only one path.

  4. 04

    Export and optionally train

    Export 14-dimensional node features, graph edges, and labels for PyTorch Geometric. Optional training supports GraphSAGE and GCN.

DRIFT sample stacked bar chart showing the share of fused signal categories in each design.
Agreement and complementary detections. The sample breakdown distinguishes signals flagged by both paths, AST alone, DFG alone, and neither. Original GitHub figure ↗

What the pipeline produces

Fused labels

combined_labels.json stores per-signal scores and fusion categories.

Readable analysis

fusion_report.txt summarizes the fused results for inspection.

Graphs for learning

Node features, edge indices, and labels are exported as NumPy arrays.

The figures are illustrative outputs from the repository’s private Trust-Hub-style benchmark corpus. The public repository does not include sample designs, so these exact plots are not reproducible out of the box.

Details and figures sourced from the project’s README and user guide.

Explore the repository ↗